Artesis eMCM misalignment case study

Industry: Chemical Manufacturing Equipment:Blower

Fault: Misalignment Nominal value: 440V, 39.1A Inverter driven

Distributor Name: TZ Inc.

Contact person: Zackai Huang

Email: [email protected]

Phone:0920-148-776

Customer had a chronical noisy blower as you could hear from below clip. (Some devices won't show the clip, if you are using computer you will see the clip)

The plant manager had used other sensors including accelerometer, their home brew award winning technology AI MCSA 1.0 where they used AI visual algorithm YOLO tried to identify the fault, the technology is developed by their company's quality department located less than 50 meters away from the blower. Due to high surrounding vibration level, they were unable to identify the root cause until they borrowed an eMCM from nearby process research and development department. The process RD was working on smart chemical plant demonstration project, where they employs Siemens WinCC and AVEVA, and perhaps they recommended Artesis technology to process RD. Minister of Economics had praised their result. However, the isolated computer that Artesis was running on had a misterious virus. While the program halt, they were able to lend the eMCM to nearby solvent plant.

Prior to installation, we went to the site and checked the following:

Plant:

The site annual produces roughly 1000tons of solvents alcohol, acetone etc, the site is a relic chemical plant inherited from the Japanese, the site was annihilated during second world war, due to the solvent could be used as fuel, and was transferred to CPC Taiwan. Currently solvent price is roughly 1 USD per KG, so the site revenue is roughly at 1 million USD.? Although the blower was running at low frequency to ensure reliability, but every year, after the overhaul, they had to restart the plant, and the motor will be running at full load for some time, and it takes a week to stabilize the process. The blower is the lung of the plant, it feeds air to burner to control the thermal energy feeding to the plants. In the case of downtime, it will stall the entire plant productions for at least 1 week. It takes a few days to fix the blower, it is estimated any downtime will stall the production for roughly 5 days, and will suffer a direct loss of 27K USD, the indirect loss is roughly 3 times, making it 80K USD. Although the motor is running at low frequency, but during their post annual overhaul startup, during their full load operation, they are very likely to encounter downtime. There is a spare motor SKU at site, the motor has fan motors to ensure adequate cooling running at low frequencies.

Asset:

Prior to installation, the following details have been checked:

Motor Type: The motor is 3 phase induction motor, driven by inverter and has a cooling motor.

Asset Type: Blower, direct drive without transmission devices, the blower has filters. No liquid coupling or hydraulic coupling present. liquids tend to act as formidable dampers.

Distance from the site: less than 100 meters for ethernet connection.

Ethernet Cable: Siemens industrial grade cable, this is the most responsive cable I have used, In Taiwan usually the cables are not suitable for eMCM requirement, we got this from the RD program Siemens buddies. Minimum ethernet cable spec: CAT 6 TIA-1005 M2I2C1E3? or ISO/IEC 11801-3

Inverter: checked the setting on the inverter, the switching or carrier frequency is at the minimum requirement 2K Hz.

Hardware Installation:

Usually the whole installation takes a few hours, mostly checking the circuit diagrams, and software installation. The eMCM is installed behind the man with blue shirt.

We installed separate panel due to insufficient room on the inverter panel.

CTs and voltage taps are behind the safety panel, we could feed 690V line directly to eMCM, with higher voltage we get it from existing CT and PTs.

The control touch panel usually installed on the panel opening pane.

Software Installation:

I prefer to run Artesis software on faster computers, so I often look for Intel i5 with 8G of ram, the size of the storage depends on scalability, there are not many motors around so I chosed 1Gb solid state hard drive. Artesis software could run on Win 2000 ~ Win 11.

There are at least five softwares to be installed: SQL express, eMCM configurator, AES, AES viewer, and SSMS (SQL management software from Microsoft), SSMS is required when the database needed to be backed up and sent to Artesis for analysis support. In Taiwan external internet is usually not allowed by companies.

Learning phase:

Due to eMCM needs to fix the frequencies for at least a few hours during learning phase, it is difficult to setup at asset startup when the process is in transient state, we had to wait for the process to stablize, but we could run it for 7 minutes to check if the asset has installation issues, I always do a 7 minute learn, and eMCM will generate automatic report and check the RPSD(Residue Power spectrum density chart, Residue is the difference with the model, we call it PSD for short) for possible installation related faults, such as misalignment, soft foot, component loose, belt issues, duct obstruction etc. eMCM is able to accurately detect existing fault prior to complete learning, learning phase will enable a more accurate detection.

The motor bearing model is often available on the motor name plate, why not throw in the numbers and check the bearing health?

We could also check the Fan blades on one click

We could also check if there are turbulent flow caused by clogged filters, if this is a pump, we could check for possible pittings on the blade that causes turbulent flow, it looks something like this (Report from other equipment):

After the learning phase, we need to run at least 10 days of learning period, to ensure the learning period covers most of the different operation windows or process conditions, and also covers Saturday and Sunday grid line, usually they are different from working days.

Automatic Report

Power quality is good, this motor has good diet for longevity.

Identify Fault:

From the automatic report, we learnt that there might be misalignment, Transmission and Unbalanced issue, component looseness these faults usually interact with one another, hence we need to look at RPSD to analyse which one.

Fault analysis:

We have identify the fault as misalignment.

Case Follow Up:

Informed Customer to check the misalignment, they did and they checked the eMCM to see if the problem is fixed.

Comparing with previous data

eMCM has the most powerful tool in the industry for analysis, the mighty Trend chart, where they isolate each fault band on RPSD, and we could see the trend of each isolated fault, we call it fault index. We know that the developing fault deteriorate at exponential rate, and exponential rate on RPSD logarithmic amplitude is a straigt line, hence if it is not a straight line, then it is not a developing fault, and most likely is operational fault, external induced fault, when it is a straght line then it is surely a developing fault, but when it is exponential on log scale axis, then we know that there are more than 1 fault stacking up. The fault bands are generated by AI, from more than 10 million motor data, each motor has an unique fault band.

Below is the Misalignment/Unbalanced fault index, the value decreased after the misalignment fix.

Below is the transmission fault index, the value decreased after the misalignment fix.

Below is the soft foot/component looseness fault index, the value decreased after the misalignment fix.

Follow up:

Customer had their improvement project on this asset, although they fixed the alignment issue, but seems they did a temporary solution, they will reduce the base by 2mm and rebore the footing holes for permanent remedies.

Why other technology failed at this site?

vibration sensors (Accelerometers) will pick up ambient vibration, but MCSA will not, for MCSA, it measures the perturbation on electrical current, the perturbation is caused by air gap displacement caused by vibrations, the displacement varies the magnetic circuit and shakes the electrical circuit, for external vibration, both the rotor and stator does a synchronized vibration, but for vibrations transmitted through the axis, it will render an asynchronized vibration between rotor and stator. Vibration sensors are great for protection system, but less so for predictive maintenance system, where the scope is focused on the asset itself. MCSA measures the vibration directly from the rotating shaft, but vibration sensors pickup the signal from the external case, the signal had to make transit from the bearing, overcome the damper guard, and walk through some part of the external case to reach the sensor.

MCSA had several improvements. The model based MCSA you see on the ISO 20958 is merely MCSA 3.0, Artesis has moved on to MCSA 4.0

As for MCSA 1.0, they usually use FFT for frequency domain analysis, there are a lot of noises, it is like looking for a metal hair pin in a rice paddy, but if you do analysis on Artesis system, it is like using a metal detector finding a metal hair pin on a cement floor.

Artesis eMCM has always find some ways to amaze me. recently we had installed on several 6 throw reciprocating compressors, eMCM was somehow smart enough to identify the reciprocating frequencies without any information given and although the signals goes beyond the alarm threshold, but it does not show as alarm on the bar chart. but you might ask what if other faults stacked behind these frequencies? don't forget that the amplitude axis is logarithmic, so when the fault is severe enough, it could still accurately estimate the life span, but I think we won't be able to detect fault at very early stage, 6 month to 1 year in advance? Who knows, maybe Artesis is working on MCSA 5.0.

Artesis has asked me to do a case study on these equipment, but due to FPG strict regulations, I am unable to get enough information for case studies. Fingers crossed that I won't get in trouble for shaing just one slide below, an healthy equipment, there is nothing wrong showing a healthy individual, I hope!