Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
VibGuy~5

Sr. Member
Registered:
Posts: 75
Reply with quote  #1 

The latest advertisements for IIot, Industry 4, Machine Learning, AI, etc. have plants populated with hundreds (or thousands) of accelerometers, all battery powered and feeding directly to the cloud via gateways. In the cloud, everything is alarmed with machine learning, so there is no need for human intervention until something goes wrong, when the cloud will send an e-mail direct to a technicians mobile to tell them they have to balance a pump or change a bearing on pump 65 out of 1000. Sound like a fantasy?

Well, the picture they are painting is that guys (or girls) doing routes are dinosaurs and are on the way out. Whether it’s true or not, they are selling a good image and in the flash bang world of Industry 4.0, image is all that matters.

I don’t agree of course, as they are discounting the 4 most important sensors (sight, touch, smell and sound-I don’t go for taste) and the sensors they are spruiking are at best, usually 5 kHz resonant frequency, 12 bit onboard A/D converter, 3200 LOR, 50 mV/g with no high frequency algorithms (no peakvue, SPM, enveloping, etc.). But, crucially, I don’t make the decisions.

What do you think? Are we dinosaurs to be replaced by Machine Learning? Should we all be doing online accountancy courses so we can put bread on the table?

OLi

Sr. Member
Registered:
Posts: 1,912
Reply with quote  #2 
You can take it easy and work hard for some time yet, nobody will pay the price. I have tested with a better system than you describe even including PV enhancements. I did a marekt research resulting in 150 chemical plants questioned here and 1 would buy something related and it was a papermill (they are getting more like chemical plants too these days) and the rest neither used vibrations or was planning to do so.... So there are a lot of work to do still but not really what I expected but the brute reality..... I did see 6 months of data from a papermachine and I was surprised of the variation. I have 20 years of route data from the same machine and I never realized the variations due to production, yes some but much more than I have seen in the route data so I got 1 case of drive system fail on a DC drive and I was like 1 hour to late as it already had blown up and stopped when I saw the data in the morning but they were surprised I would know.... So you need more sensors to make sense of the data if you are not on site....... 
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Alex

Avatar / Picture

Sr. Member
Registered:
Posts: 204
Reply with quote  #3 
Industry 4.0 sounds realy exciting to some managers. Like going from WIN7 to WIN10. The most interesting part is some companies would jump from the state without ANY measurements to online measurements. Here is one half real example: 70 middle size fans. Online monitoring investment costs: 300.000 EUR. After some time, OMG what about calibration, we forgot that part. We need to recalibrate every sensor after 2 years. It's better to have reserve set of sensors = + 200.000 EUR or we will stay without online monitoring for some time. Now we have to pay 30.000 EUR for annual maintenance, we didn't count on that part. And than there is a red alarm on one of the fans. What is going on? We have no idea. We need to call an expert. Yearly costs 30.000 EUR. It would be better to have educated personel by ourselves. Let's send 2 people on intensive course 10.000 EUR. We had no luck, one of them went to another company, the other is not interested in the job. And then suddenly one fan broke without the alarm. How is that possible? ISO standard table says it should work but it doesn't any more. What now. We go back to what was suggested in the first place, but we wouldn't listen: periodic monitoring. [smile]

For that money they could buy every new fan in 4-5 years!

P.S.: They all forget nothing lasts forever so does not online monitoring equipment.
weller

Member
Registered:
Posts: 55
Reply with quote  #4 
I somewhat agree,plenty of hype and B.S. to go around but we have found the wireless /battery accels have their place. We use both route and near continuous monitoring. A complex gearbox would be analyzed with a 2140. Constant speed pumps and fans are easily watched by wireless by establishing trend data. The 70 fan example would be closer to 150k (US) install price for the type we use in plant. Approximately 20k to monitor per year. Variable speed /load is trickier. In this application usually other  data is necessary to overlay with the vibration data.When overlay-ed this also opens up the question of making changes in the process. Obviously a remote analyst cant make that decision alone. In house example: center-less grinders with high regulating head failure rate. By adding near continuous monitoring and component failure analysis, failure rate went from 30 days to 9 months. On the process side, operator set-up can be verified by a certain frequency signature. I strongly agree with educating in-house personnel and a mixture of vibration collecting practices. Experienced people on site make all the difference. Cooperation between process , engineer and the analyst is another key component. All that said we're still taking baby-steps, a lot to learn.
Shurafa2

Sr. Member
Registered:
Posts: 170
Reply with quote  #5 
Do you know industrial plants that have fully implemented IIoT, Big Data, Machine Learning etc?

This is a question I keep asking when I attended events and this topic is marked. I hear typical answers: yes, we WILL work will plan x, company y PILOTED that system, facility z showed interest and conducted an assessment of the implementation etc etc.

Unfortunately, it has been overmarketed, overpromised and many times oversimplified to the customers. At least in the heavy industry like chemicals, petrochemicals, power, oil and gas, there are infrastructural limitations to begin with. Many times, some basics are not available. I'm not aware of other industries but these industries are "slow" in nature to modernize.

However, IR4 is coming sooner in some places than others. Till it comes to where I work, I'll need to visit the machines, listen to them, touch them and read the old gauges.

Regards- Ali M. Al-Shurafa
JuddJones

Sr. Member
Registered:
Posts: 382
Reply with quote  #6 
I know of one company that is making a pretty good run at installing IIOT wireless equipment nationwide at one of the biggest companies in the world. It is pretty early, we will see how they fair. I am keeping an eye on it. I think 20 years from now it will be the standard. We are still too early. It will not be long before we will see wireless Mems accelerometers with frequency response equal to the best we are using with order tracking capabilities at a fraction of the cost of ICP accels. We have 2 different wireless systems here that we are playing with at our plant. If they could work on equipment with VFD's we would make a significant investment right now.
Sinski

Avatar / Picture

Sr. Member
Registered:
Posts: 394
Reply with quote  #7 
We were meant to be getting some of these wireless sensors for a trial. That was over a year ago I heard that and still nothing. This was the idea of a national engineering manager who has nothing to do with the vibration side of things.

This is how they will market and sell. They do not want vibe analysts making decisions on these things. They will market and sell straight to managers as that will be easier to say it will solve all of your problems to them. Its a bit like how SAP would sell their software to companies by going way above the level of people who need it and tell them it does everything when it really doesn't.
OLi

Sr. Member
Registered:
Posts: 1,912
Reply with quote  #8 
Well that's why I have a hard time selling them, all managers I used to know have retired....
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
VibGuy~5

Sr. Member
Registered:
Posts: 75
Reply with quote  #9 

I’ve heard of one company installing 1100 sensors across their site and discontinuing the route system. They’ve gone for the cheaper MEMS sensor with temperature as well, on the basis that it might have a lower resonant frequency, but the temperature will tell them if something is happening-I’m not sure if that’s a marketing strategy.

I know the MEMS sensors are increasing their capabilities every year, but of all the wireless providers I’ve looked at, nobody has any equivalent of peakvue/shock pulse/enveloping. I agree, they’re probably great for constant speed fans and pumps, but not for critical plant.

OLi

Sr. Member
Registered:
Posts: 1,912
Reply with quote  #10 
Vendor
These have it, it is made by the last/latest guy working on PV development in Knoxville after his early retirement .....
https://www.icareweb.com/en-gb/monitoring-tools/i-care-products/

I don't see a MEMS explosion, there are only 1 raw sensor supplier I know of and they don't know what they are being used on and the pricing are still after near 10 years not competing either with shear piezo or MCV at the same spec level and volume. I don't see adapted airbag sensors that are lo freq a real option, but that's me. I still wait for a good MEMS that I can use to a competitive price.


__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
GaryVibe

Member
Registered:
Posts: 16
Reply with quote  #11 
I have experienced similar issues. A NON Vibration Trained individual was working with a team to propose wireless accels on a pump. I, by chance was invited to a meeting to highlight the technology.  What I found was extremely disappointing. Three axis mems sensors mounted inside a plastic box with a 2 leg magnet placed on a pump motor running at 3600 rpm. The sensors had a 2 Hz to 1,000 Hz response range with about 800 LOR. 
I asked about being able to see vane pass at 1400-1500 Hz and electric motor issues, along with the usual 1st stage bearing data at 1,000-1,500 Hz on this machine......You would have thought I had turned into a 3 Headed Monster that just walked into the room. Not one person on the "Team" understood what I had just asked and they are trying to sell me a wireless product to monitor the health of my equipment. 
I am all onboard with remote monitoring, but you had better provide me with adequate ranges and resolution to give me a sense of change in the machine health that is important to me, and not your sales department. 
weller

Member
Registered:
Posts: 55
Reply with quote  #12 
Do you know industrial plants that have fully implemented IIoT, Big Data, Machine Learning etc?

I dont know of anyone "fully" IOT, but i do know people that have used it  in areas of production.Other examples outside my personal knowledge are all over the internet. Yes it is hyped and oversimplified.At our plant we are sticking our toe in the water but what is needed is a dedicated Data Scientist, which we are not. ( My personal goal is to learn machine learning coding. I have the electrical/mechanical background but the other side of the coin is the coding.) IOT is just another strategy so implementing a good technology or good strategy without the proper people or skill set will fail. Certain processes in our plant already have data historians for troubleshooting purposes. The historian is usually only looked at when there is a problem. A data scientist trained in the particular process can develop algorithms and models that could improve the process and predict failure. Installing wireless aceels is not really IOT, it's just one small part of the whole. Any complex problem usually requires a complex solution which ultimately depends on good people working together. This also applies to politics!
Shurafa2

Sr. Member
Registered:
Posts: 170
Reply with quote  #13 

There could be some plants that have really done it. But they are so few and not proportional to the amount of "propaganda" on the net and conferences.

Plus, there are conflicting or confusing messages to the end users. Based on the current level of maturity of the technology, what kind of existing plants are the best candidate for this revolution?

If my failures are in general slow in nature like unbalance, misalignment, bearing faults etc how much additional sensors will be useful? Must they be wireless? Is it necessary send the data to the cloud? Do my staff know the basics of diagnosis so they train the machine and let it learn?

What is presented in the events and on the net is usually the bright side. How many attempts unfortunately failed and went unreported? I can tell you, many. And how many simple yet powerful improvements in the classic systems (with less fancy words) are not highlighted? Is sending an alert email from the historian system considered a part of this new trend?

Sorry for the so many questions.

Regards- Ali M. Al-Shurafa

Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.