I am interested in hearing from folks who have experience in either striving for or making a significant impact to their ratio of Reactive/Preventative/Predictive work order ratios based on the use of predictive technologies and RCM approaches specifically on diesel power plant applications.

At Bermuda Electric Light Co. Ltd. we are exploring RCM within the context of a predominantly diesel plant with a peak demand in the region of 120MW to date and growing.

I'm striving for that cost- and time-saving measure through a hardware/software product I'm helping to develop, so I'm on the other side of the usage scenario. We're developing diagnostics for diesel engine power generation plants using a combintation of performance measurements and feature extraction from typically-monitored signals. We're in the beta-test site stage, so maybe there can be some mutual benefits.

Best regards,
Jeff

This message has been edited. Last edited by: MJroc,

Abayomi,

I’ve worked with marine diesels for over 30 years, both propulsion units and power generation and the key as you know is doing the right maintenance at the right time. I saw the trend go from more time directed tasks to more condition based or need based maintenance. The ROI for sensors and diagnostic equipment paid for itself in increased engine availability and focused maintenance periods. Once condition monitoring was established maintenance became focused on actual problems, single cylinder overhauls rather than wholesale engine overhauls. We monitored and trended the key performance indicators of the engines, figured out what parameters we were not monitoring that we should be, gathered the technicians and identified those time directed tasks that could be modified.
A basic plan I would consider:
1. Get your organization trained from the top down on RCM. You absolutely must have top management buy in to the program.
2. Evaluate the current maintenance plan for effectiveness.
3. Change from Time Directed (calendar or hour meter) to Condition Based maintenance.
4. Invest in a good oil analysis program
5. Look at getting either online or periodic Diesel engine analysis, Dynalco makes a very good analyzer and I’m sure there are probably others out there. Current technology allows you to get a very good picture of what is going on inside the engine. Simple things like using airborne Ultrasonic’s to listen to the valve train and using an infrared thermometer to check fuel injection line temperatures will pay immediate dividends once the maintenance technicians get used to using them.
6. Evaluate what conditions you already have monitored that you can get historical readings on and trend. I’ll bet if you start trending readings you already have, oil pressures, turbocharger and cylinder temps and compare to past maintenance records you will start to trends emerge.
I’ve concentrated on the engine side of the plant but the same holds true for the generators. Using airborne UT and vibration analysis to check bearings, infrared camera for tracking the generator heat signature are good places to start.

You may see an initial increase in problems identified, but I attribute those to being hidden problems you were not aware of that you can now schedule and plan for, less reactive more planned.
Eventually if you keep working the plan you’ll diagnose problems earlier and tweak the maintenance practices to prevent any of the obvious avoidable ones and expand the use of your pdm program as you gain confidence in it.

http://www.dynalco.com/
No I don't work for them they just have a good product.

Regards
Hank

Jeff
I would be interested to hear more about your project. In terms of technologies with potential to monitor diesels for failure modes I am aware of:

Engine Diagnostics: Assessing cylinder component operation, timing and condition via cylinder pressure and its derivatives vs. crank angle
Oil Analysis: detect component wear
Ultrasonics: may be limited to exhaust leaks though other areas may be experimental
Temperature measurements: via thermocouples to indicate bearing issues
Oil Mist and Splash oil monitors: to indicate bearing issues (earlier)
Monitoring for Crankshaft Bearing defects via electrical current based system (fairly new to the market, said to be even earlier than splash oil)

It sounds as though you software aims to process (temperature/vibration/pressure?)data into what would apply to some sort of PF curve?

Hank,
This is very helpful. Thanks. What size engines were you working with? We are up to 14.5 MW. I will check dynalco out.

The project used some of the techniques you mentioned above to assess performance and component degradation. The engine size was smaller; however, I think the principles would still apply to larger machines.

Basically, I would come at it from a couple different angles. We characterized normal pressure profiles and detected excursions from them using an empirical model. So it baselined operation, and we were able to detect early deviations from it. But, on other machines we've been able to diagnose bearing faults very early, as well, using accelerometers and feature extraction. If you baseline the performance-related parameters using a model, and also capture some vibration features, it could form a good picture of overall operation.

Would you be interestd in being a Beta site for something like this?

Jeff

Jeff,
I'm not familiar with the requirements and/or implications of being a beta site and I'm interested to hear more. Feel free to email offline if needed.

I am also looking at the dynalco equipment which certainly appears comprehensive. We probably have most of the equipment separately but this integrates them all relative to crank angle/valve timing etc. Hank: Did you get significant formal training of this?

I don't have your email, so I'll describe what I see the Beta site arrangement being like (I'm flexible), and you can follow up with me if you're interested.

We could install our software and hardware at your site at a reduced cost in exchange for feedback on the development. We've proven that our approach works (I can give you more details offline), but are interested in getting more user feedback to help us crossover from development to usage. In weighing your pros and cons against going with someone like Dynalco (maybe even in parallel), we can afford the customization that you might need because we have a vested interest in your liking what you get. Not that they don't; we are just at a position where some customization is expected because we don't have a solid footing yet. But, because it's not 'out-of-the-box' ready, you're undertaking some risk, so it's only reasonable to expect to pay less. Regarding integration, if we have access to your existing equipment signals, I assure you that we can assess them as a whole.

If you'd like to talk more, email me at:
kerjef {-at-} gmail {dot} com

Anybody knows these new condition monitoring technologies for rotating equipment:
a) Electrostatic engine monitoring
b) Automatic diagnosis techniques based on
neural networks, expert systems and
statistical methods.

No, but if you have some insights, we'll be more than happy to listen...I have heard of all these techniques but not applied to diesel engines in particular.

Separately, I am hearing positive comments on Schaller's Becoms failure detection system. I'm told the system detects metal to metal contact throughout diesel engines and has been refined to locate a specific chamber or area where the fault is developing. It is said to me much faster than oil mist or even splash oil systems.

Does anyone have experience with this or a similar system?

Read more about the Schaller systems here:

http://www.engineeringtalk.com/news/scv/scv102.html

and here

http://www.schaller-gmbh.com/

Josh-

I can speak to the second one you mentioned, automated diagnostic techniques, but not the first one. I've actually spoke with Abayomi a bit about them, too; but the company I'm with has successfully applied them to several systems.

I think they're great techniques if you have a good set of data to provide to it. For example, I've helped develop a piece of software that determines what type of fault is occurring in a gas turbine engine in a production facility test cell. The intent of the software is to reduce troubleshooting time by looking at patterns in the data to determine what specific fault is going on, and therefore directing the maintainer (in this case, test cell technician) to the specific engine component causing the loss in engine output so that it can be repaired. It used a combination of pattern recognition and statistics. It works great; but we had access to a numerical model from which we could get the signal patterns characteristic of specific faults.

There are several packages out there that can help in applying neural networks (or really any of the automated diagnostics) techniques. Matlab has a toolbox for it; Neurosolutions is well-known, etc. Even some open-source things will work, too. The beauty of neural networks is that they're just a large network of simple functions (such as sigmoid activation functions), so they're not too complex to program. The difficulty comes in having a good, clean data set to train on; and in having expertise to pragmatically apply the results--the 'whole package' so to speak.

I'd be happy to discuss it more, if you like, whether you have questions about their application in general or if you have a specific request.

Regards,
Jeff

Abayomi, I recently joined the forum and saw your post. My company has been supplying diesel health monitoring systems to the US Navy for over ten years. Diagnsotics and fault prediction (prognostics) are automated through software. Please visit www.macsea.com and www.macsea.com/agents for some background and contact me if interested.

Sincerely,
Kevin Logan

Kevin,
Thanks for the links. I have had a quick look and downloaded the demos. Can you comment on any additional equipment required to interface with this software?

On your success with this product?

Kevin, what would be the benefits of applying this DEXTER technology for diesel engine-driven fire water pump & emergency generator which are not continuously run but we have to test run these equipment biweekly at the moment?

Good Day to all, short info to the Diesel Engine Analyzer from Dynalco.
We have used the Recip-Trap from Dynalco, its a good system with many options, but heavy to carry and difficult to handle with all the cable and accessories and the price .......
We use now the analyzer from IMES ( www.imes.de )- check it out -, very light, easy to handle, has not so many options like Recip-Trap, but low in price, pressure sensors are high temperature and can be used also for online monitoring ( also available )

Regards

Hello Hank Cutterman,

Based on your quote:

quote:

Originally posted by Cutterman:
Simple things like using airborne Ultrasonic’s to listen to the valve train and using an infrared thermometer to check fuel injection line temperatures will pay immediate dividends once the maintenance technicians get used to using them.

Can you please tell me the significance of fuel injection line temperatures?

Regards
Ibrahim

In addition to Dynalco, here are a couple of others who specialize in recipricating engines.

www.windrock.com

www.prognost.com

Both of these companies measure engine performance as well as condition monitoring.
I have no affiliation with either company.

Dave,

I had used an earlier version of Dynalco's Diesel-Trap and recently used Windrock system. My favorite still remains as Diesel-Trap because it is very easy to use and detect problems with. To anyone interested, I would recommend Dynalco.

However, both Dynalco and Windrock are very much overpriced if you ask me. Anything initially designed for the nuclear industry seems to carry a hefty price tag.

I am curious when we will see standard data collector manufacturers display time domain data with crank angle on the x-axis instead of time. Throw in a pressure transducer (which is possible with all data collectors now) and get P-V diagrams, you are halfway there. (Surely it is not that simple but ...)

Regards
Ibrahim

PS: I have no affiliation with Dynalco

Thanks Sven/VibraTek. Can you give us some examples of what has changed in terms of your maintenance using this equipment? Any significant findings, failures avoided or extensions between outages?

Dave G. Thanks for the links. I will check these out.