Jump to content
Strawberry Orange Banana Lime Leaf Slate Sky Blueberry Grape Watermelon Chocolate Marble
Strawberry Orange Banana Lime Leaf Slate Sky Blueberry Grape Watermelon Chocolate Marble
Cpickle

How much Solar v Batteries? What do you have?

Featured Posts

1 minute ago, Tony Brooks said:

That will pay for a lot of batteries. On present form I could buy three sets for less and I expect them to last 5 years each set. I will be dead before the payback time is up!!! Or at least no longer able to boat.

Absolutely, and that's why it's aimed squarely at the military. It's used by Enersys (the biggest battery manufacturer on the planet), the UK MoD, the US DoD, the UAE MoD, General Dynamics, Dytecna, Thales, BAE systems and NP Aerospace. They have all tested it, all approved it, and all fit it as standard and no longer bother even auditioning anything else.

It's got to be pretty good ;)

Share this post


Link to post
Share on other sites
3 minutes ago, Dr Bob said:

I was thinking about a group of guys doing it for free, ie like CanalPlan

They wouldn't have the knowledge. 

Share this post


Link to post
Share on other sites
6 minutes ago, Dr Bob said:

No, I was thinking about a group of guys doing it for free, ie like CanalPlan (which I find wonderful) - collecting data sets and seeing how to predict. I only managed a modelling team, rather than did the nitty gritty of the modelling so wouldnt be capable of the model crunching but defining parameters, data collection and model validation are the time consuming things which is within the scope of many of us.

How would that work, all the data collection in the world won't tell you the SoC unless you had something to accurately get the SoC for the data to model against.

Share this post


Link to post
Share on other sites
3 minutes ago, Robbo said:

How would that work, all the data collection in the world won't tell you the SoC unless you had something to accurately get the SoC for the data to model against.

Yes, of course all data sets would have to have a reasonably accurate SoC values and Capacity numbers as I guess they may be the target outputs. I've not thought it through but for each system modelled you would have to log a SoC - maybe via SG - but it could be done via voltage at rest and linked to voltage under load. The trick is to collect a lot of data so that can then be assessed to reject poor data and to get good averages so you dont have to rely on one off numbers that could be inaccurate. Time consuming but not difficult.

Share this post


Link to post
Share on other sites
35 minutes ago, Dr Bob said:

Yes, of course all data sets would have to have a reasonably accurate SoC values and Capacity numbers as I guess they may be the target outputs. I've not thought it through but for each system modelled you would have to log a SoC - maybe via SG - but it could be done via voltage at rest and linked to voltage under load. The trick is to collect a lot of data so that can then be assessed to reject poor data and to get good averages so you dont have to rely on one off numbers that could be inaccurate. Time consuming but not difficult.

Unless you have a consistent method that all use to accurately get the SoC the data would be useless.

Share this post


Link to post
Share on other sites
10 minutes ago, Robbo said:

Unless you have a consistent method that all use to accurately get the SoC the data would be useless.

Fully agree. I think you would have to be consistent across all methods of data collection.

Share this post


Link to post
Share on other sites

Gibbo would tell you how easy it is. Test literally thousands of batteries to destruction, cycling them to various SoC until they die, charging them at different voltages, taking relative density readings at all stages, and collating all of the information to arrive at an algorithm. Write a program to make the results self-correcting and put it all in a box for less than a hundred quid. 

Dead simple. 

Share this post


Link to post
Share on other sites
3 minutes ago, Dr Bob said:

Fully agree. I think you would have to be consistent across all methods of data collection.

You could release the software for the actual device tho so people can use and come up with there own figures for the algorithm.

Edited by Robbo

Share this post


Link to post
Share on other sites
40 minutes ago, WotEver said:

Gibbo would tell you how easy it is. Test literally thousands of batteries to destruction, cycling them to various SoC until they die, charging them at different voltages, taking relative density readings at all stages, and collating all of the information to arrive at an algorithm. Write a program to make the results self-correcting and put it all in a box for less than a hundred quid. 

Dead simple. 

Yeah, a soldering iron and a few odds and ends from Maplin, job done. Don't know what the fuss is all all about, money for old rope. 

Share this post


Link to post
Share on other sites
3 minutes ago, WotEver said:

Gibbo would tell you how easy it is. Test literally thousands of batteries to destruction, cycling them to various SoC until they die, charging them at different voltages, taking relative density readings at all stages, and collating all of the information to arrive at an algorithm. Write a program to make the results self-correcting and put it all in a box for less than a hundred quid. 

Dead simple. 

I wasnt thinking of testing to destruction....and I wasnt thinking of testing thousands of batteries....just a couple.

I also wasnt thinking of an algorithm. For that you need lots of data - and using thousands of batteries - as his model was a global model that had to work for everyone. I was thinking of the newer types of model predictions - viz Topology that uses pattern matching or other advanced techniques. The technique is to get one or two models working on one or two installations during normal running with a limited set of variable inputs and a defined set of fixed inputs. If a relationship can be set up between data in one set up, then the model can be expanded one by one to introduce new training sets. The world has moved on. The amount of data you need is far less.

Please stop reading now if this is boring you.

Let me give you an example of pattern matching models that many of you may have an interest in.

The refining industry make gasoline and sell it to us motorists. They sell to a standard MON/RON value (Motor Octane Number, Research Octane Number). So I buy petrol at RON95. This is tested by an engine test which takes 24 hrs and is accurate to +/- 1 unit. The refineries therefore should make it made to RON96 so that every drop is over RON95 when they sell it ( I think I have RON and MON the right way round) - similar to shops selling a bag of sugar of 1Kg, they always put a little more in to keep them at 1Kg minimum. That 1 unit of RON 'give away' costs refineries over £2M per year. We developed a model to do this on line using Near Infrared Technology (NIR). At one refinery we tested the output from gasoline blending to collect NIR spectra - that were to the naked eye just a wiggley line - and tested a number of Engine tests which were inherently inaccurate.

People had tried for years to do this via regression modelling and trying to come up with Algorithms to do it. They never worked. The input data kept changing and the linear regression models although accurate when the new feed was in the training set range, were useless when outside the range. That is akin to having a model based on say a 500Ahr battery bank and then using the model on a 400Ahr bank.

The topology method overcame this and it was possible to come up with a unique system that could cover all feeds with only a little effort with new tests and model validation.

Now when you go to a petrol station, you tend to get 1unit of RON less as there is no give away as the NIR test is accurate to within .05% rather than the 1% of the engine test. If you are into Diesel not petrol, then we did the same for Cetane number in Diesel. Have you ever noticed how sometimes you get a tank of bad petrol (or diesel). Sometimes it is this RON giveaway. Sometimes it is how much Butane the refinery puts in the petrol!!! ....but not all refineries are using this technology - we were selling this technology at £500K a pop.

Maybe Gibbo is using advanced techniques but pattern matching and the newer techniques (ie sparse data) dont require you to determine an algorithm - and the use of that word suggests it is regression (or regression type) technology. The new methods work by collecting live data but in this case you would also need to get some SoC data (but not necessarliy accurate - as long as the method is consistent) and capacity data.

If you wanted to sell such a model then YES you would have to put so much effort in to make it work for everyone that you would be selling at £1K a pop. I was thinking about a model that works for one person, and then training it with data for another etc etc. Data logging is perhaps the biggest challenge but perhaps linking a Raspberry Pie to a Voltage and Ahr counter could be a cheap option. There would be a lot of other inputs to measure as well (engine revs, temp, solar input to MPPT, output of MPPT etc).

It is nice to dream........ this is more a hobby for a bored ditchcrawler......and I am not bored anymore!

Share this post


Link to post
Share on other sites
20 hours ago, WotEver said:

I had a look on a site for gels as well, it seemed to indicate that an alternator was ok but for solar I did say buy a quality controller, like in my case a midnite which can be set up for gel easily. For me I think they will be ok for the electric bathtub until sodium batteries become available

Share this post


Link to post
Share on other sites
52 minutes ago, peterboat said:

I had a look on a site for gels as well, it seemed to indicate that an alternator was ok but for solar I did say buy a quality controller, like in my case a midnite which can be set up for gel easily. For me I think they will be ok for the electric bathtub until sodium batteries become available

Great - a boat on water, in the rain and sodium! I also understand they need to run at high temperatures.

I think I will stick with relatively safe lead acids for a long time yet!

Share this post


Link to post
Share on other sites
2 hours ago, Tony Brooks said:

Great - a boat on water, in the rain and sodium! I also understand they need to run at high temperatures.

I think I will stick with relatively safe lead acids for a long time yet!

When I was looking at the sites they made no mention of high temps in fact that is the plus for them they operate at either end of the scale unlike lithium which seems to like catching fire will have a look at the inventors site later

Share this post


Link to post
Share on other sites
10 minutes ago, peterboat said:

will have a look at the inventors site later

Presumably you mean one of the developers' site? Sodium batteries were first played around with in 1981, and it's only recently (last couple of years) that some developers have looked more closely at them. 

Share this post


Link to post
Share on other sites
25 minutes ago, Tony Brooks said:

The ones I knew a very little about used liquid sodium so it had to be hot.

Yeah, but they're playing about with all sorts of different compounds at present.

The obvious advantage of them is that unlike lithium, sodium is available in abundance. However the three current disadvantages are reliability (which is being addressed with different (solid) electrolytes), low energy density, and short life.

There are a few groups researching them and filing patents every few minutes so if they're going to prove to be viable I'd think we'd see the first commercial applications within another year or two. 

Share this post


Link to post
Share on other sites
On 20/07/2017 at 20:15, WotEver said:

 

The obvious advantage of them is that unlike lithium, sodium is available in abundance. However the three current disadvantages are reliability (which is being addressed with different (solid) electrolytes), low energy density, and short life.

 

You didn't mention the biggest problem with sodium.

It makes your blood pressure too high apparently, and my doctor says I must stop eating it. The idiot. 

He'll be telling me alcohol is bad for me next. As IF. Huh.

Share this post


Link to post
Share on other sites
1 minute ago, Mike the Boilerman said:

 

You didn't mention the biggest problem with sodium.

It makes your blood pressure too high apparently, and my doctor says I must stop eating it. The idiot. 

He'll be telling me alcohol is bad for me next. As IF. Huh.

Best advice for a long and healthy life? Stay away from doctors!

Did you hear about the old boy, passed away at 95. He put his longevity down to eating a teaspoonful of dynamite every day. Boy, his cremation went off with a bang. 

Share this post


Link to post
Share on other sites
On 20/07/2017 at 20:26, WotEver said:

Best advice for a long and healthy life? Stay away from doctors!

Did you hear about the old boy, passed away at 95. He put his longevity down to eating a teaspoonful of dynamite every day. Boy, his cremation went off with a bang. 

 

Did you hear about the copper who arrested two lads for eating gunpowder and drinking battery acid...??

 

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


  • Recently Browsing   0 members

    No registered users viewing this page.

×