AEVA Members: your Forum Login (here) is NOT the same as your new AEVA member login.
You do not need to change your existing Forum login.

Interworking a Sunny Island with a Redflow Flow battery

non-EV Solar, Wind and other renewable power sources
non-EV batteries and other energy storage stuff
Forum rules
Important!
This forum is for discussion of Non-EV matters.
Post Reply
hyphenated
Noobie
Posts: 5
Joined: Fri, 27 Dec 2019, 09:30
Real Name: Neil Carpenter

Interworking a Sunny Island with a Redflow Flow battery

Post by hyphenated »

So I have an overall outline design of a hardware interface box that might do the job of matching a SMA inverter/charger to a flow battery. So why am I waving it around? Because I am a light electronics and telecommunications engineer, and this is 48 to 60VDC with potentially hundreds of Amps and perhaps 10F of capacitance (ie thousands of Amps). If I stuffed up a breadboard circuit, the quantity of magic smoke was small. Here it is potentially a lot more, and I’d appreciate some words of wisdom from the experienced. :D

Back in the antediluvian past (around 4 to 5 years ago) I was attempting to purchase and install one or two Redflow flow batteries to perform daily store and discharge of PV energy to cover our overnight refrigeration demand. I had some email communication with various Redflow personnel at the time. Things didn’t progress for various reasons, and a few years ago we shut the business down. Our Ergon bills are still ridiculous, despite having 15kW of inverters and double that in panels. The harsh reality is that batteries are unlikely to pay their way, and the even harsher reality is that flow batteries are a great deal more expensive than lithium. However, as an engineer I find the use of Lithium chemistry for a daily charge and deep discharge absolutely idiotic, and am prepared to pay a premium to get something with at least ten years of functionality. I’ve put aside some mad money to get a pair of flow batteries.
I know that the SMA Sunny Island did not play well with Redflows, but unfortunately we have considerable sunk cost in SMA, including the Home Manager, so despite Victron being superior when considered in isolation, I have purchased a Sunny Island 8.0-13.

I am stupid enough (or confident enough – you choose) to think that, despite Redflow having failed to achieve good results with SMA Sunny Islands, there is the possibility of a better result five years on.

I have had some discussions with Redflow, and to summarise the major sticking point, the Sunny Island throws its toys out of the pram when a connected battery stops behaving like a battery. If both ZBMs are fully charged or going through a maintenance cycle or discharged, they disconnect from the DC Bus. Despite the BMS telling the SI over CANBus that there is zero permitted charging, there appear to be edge cases where either the SI is dumping energy to the bus, or fails to see a voltage on the bus; which causes the SI to go into hard shutdown after a few minutes.
    • 1. My thought at the time and now is that a small auxiliary 40+V power supply fed to the bus through an isolating diode, and with either foldback or current limitation would have just enough grunt to spoof the SI. This would need to be associated with SOC information fed to the SI, and the SI set to disable startup until the SOC was above a threshold.
      2. The SI has hard SOC limits designed to avoid destruction of LiIon and Lead Acid cells, which would prevent complete use of the ZBM energy. The obvious fix here is a manipulation of the transferred SOC information, although if a pair of ZBMs is operating in a staggered manner, this would only be needed when a single battery is running.
      3. If the SI does put energy on the bus, we need somewhere for it to go if the batteries are unavailable, until the SI catches up with the instructions. Putting a supercapacitor in parallel with the bus would provide an energy sink.
      4. If the bus voltage spikes, there needs to be an energy dump that prevents overvoltage or high energy spikes. Putting a relay or SSR-switched resistive dump load tripped by an overvoltage sensor would fix that; and appropriate transient suppression diodes would address spikes.
So throwing all the upstream protection ideas into the blender and hitting frappé, we get the Frankensteinian Protection System (FPS):
  • • Use the BMS to trigger a dry contact relay when both batteries are offline, whether at full charge or stripping, and send a zero permitted charge instruction to the SI.
    • 5F 64V supercap string (4 off 20F 16.8V modules) precharged to 40V by diode-isolated low-current power pack. Let’s assume the labels lie; the string would be a couple of Farads.
    • 60V TPD (transient protection diode) across string.
    • The string has a voltage sensor/timer on it with a trip setting of 57V, timer set to appropriate value to reduce charge without burning out the dump resistor.
    • Voltage sensor has a relay contact to a large DC relay or a non-fake SSR with a dump resistor load (400W 4 Ohm encapsulated resistor).
The FPS presents 40 to 48V to the Sunny Island. If the SI throws a cog and commences charging, it will take around 20 seconds at 50A to reach 57V from 40V, which is hopefully more than enough time for the SI to wake up and read the registers saying 100% SOC and instructing it to stop. If not, the relay will close, dumping some of the supercapacitor charge into the resistor. Either the timing is tweaked to keep the capacitor voltage in range (ie above 40V) or use a second voltage threshold is used to open the relay again. So in theory the protection circuit can keep on trucking indefinitely.

The TPD is to absorb high energy low duration spikes, might put another one across the capacitor and across the ZBM DC bus.
As the saying goes, the devil is in the detail.
  • • A kosher supercapacitor bank of 60V would cost thousands.
    • Going cheap Chinese is fiscally possible, whilst trying to avoid the outright fake (relabelled rejects). I’m not too worried about high leakage or lower capacitance, as long as the cells are similar.
    • To get to a reasonable voltage rating, several six-capacitor modules will have to be run in series.
    • Cell balancing is often only done when approaching or exceeding the maximum cell voltage.
    • Individual cells may get driven negative.
    • I can add two layers of additional protection – Schottkies across each cell to limit the degree of negative voltage build-up; High-amperage blue LEDs across the cell to provide another cell overvoltage protection path; and might put some small LEDs in the protection circuit to show when it is active.
    • Active charge balancing is possible, but costs more than the cells.
    • It is pretty obvious, however, that high-current applied to a 24-cell supercapacitor string is going to overwhelm any cell protection.
So my major concerns:
  • 1. Keeping the supercapacitor string within some form of balance by limiting charge inflow;
    2. Fail-safe construction, wiring and fusing;
    3. Keeping current flows in general under control;
    4. Providing sufficient isolation points;
    5. Performing some voltage and current logging to see what goes on, and hopefully refine the solution (rather than seeing it fail miserably).
    6. Sufficient physical isolation from the environment and other equipment.
Any thoughts? Thanks!

paulvk
Groupie
Posts: 245
Joined: Fri, 23 Oct 2015, 23:45
Real Name: Paul
Location: Sydney

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by paulvk »

Your looking at the similar problem as I have.
I have a 200Ah nickel iron battery which needs 1.65 volts per cell to charge but is 1.2 volt per cell discharge.
With my inverters 50 volts is a good battery voltage so 41 cells gives 67.65 volts charge max of inverter is 60 volts.
The flow battery is similar.
So I am experimenting with my lead acid bank and the nickel iron, at present just 36 cells across the lead acid and I find they are providing around 20% of the power.
After new year I am going to try DC/DC boost inverters to convert the lead acid voltages to the higher nickel iron voltages,
I will keep the lead acid on but switch the NiFe bank to the DC/DC boost charge.
You may want to look at NiFe as their life is measured in decades not cycles and can take abuse over charging just cases gassing.
There are some inverters able to cope with NiFe voltages.

Regards Paul

hyphenated
Noobie
Posts: 5
Joined: Fri, 27 Dec 2019, 09:30
Real Name: Neil Carpenter

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by hyphenated »

Hi Paul

The Redflow needs 57V (absolute maximum charge rate), 56V normal for charging (I believe), which is well within the Sunny Island capabilities, assuming the charging voltage is your issue. Redflow is not the fastest in or out. 42.2 out and 67.65 in for your cells *is* a wide gap. I'm more worried about preventing the SI from a shutdown.

Cheers
Neil

paulvk
Groupie
Posts: 245
Joined: Fri, 23 Oct 2015, 23:45
Real Name: Paul
Location: Sydney

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by paulvk »

preventing the SI from a shutdown
Yes and maybe the NiFe could do the job as they would work as the dump load voltage regulator and unlike lead acid will not complain
the 1.65V is for a full absorption charge. With the 36 cells I have across the lead acid they are charging at 20 amps when the
lead acid are at 56.7V in the absorption cycle I need to get the voltage up for 41 cells to use them as my main battery
the lead acid will be there to prevent my inverters from shut down.
the only thing is they are wet cells and need some maintenance
Redflow is not the fastest in or out
The NiFe are the same 50 amps is about the most for 200Ah cell.

User avatar
weber
Site Admin
Posts: 2746
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by weber »

@hyphenated Welcome to the forum. Can you explain why you say, "I find the use of Lithium chemistry for a daily charge and deep discharge absolutely idiotic, and am prepared to pay a premium to get something with at least ten years of functionality."

There are 5 or so different lithium chemistries readily available, e.g. LCO, NCA, NMC, LFP, LTO. The last 3 seem to me to be well suited to daily charge and deep discharge (to 70% or 80% DoD), and some manufacturers provide 10 year warranties (with a pro-rata component).
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

antiscab
Senior Member
Posts: 2601
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by antiscab »

My lifepo4 batteries turn 10 this year, still in daily use. Getting to 10 years isn't so hard now.

OP - compared to cost of a super cap, have you considered having a small buffer battery instead? 100Ah at 48v would cover it
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells

hyphenated
Noobie
Posts: 5
Joined: Fri, 27 Dec 2019, 09:30
Real Name: Neil Carpenter

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by hyphenated »

Hi Weber - thanks for the welcome.

I will confess I am not up on the current Lithium battery variations, and let me be clear - for EV applications currently Lithium seems to have most of the cards, if only because of density and development maturity. But the three main issues for me in a static environment in the subtropics are cyclical degradation, heat derating and fire risk. So, as a thought experiment, what is the expected percentage deterioration for the various lithium chemistries when operating daily charge/discharge and operating in a 25 to 35 degree high-humidity environment?

I figure I need 20 kWh for day/night energy transfer, and a little more would always be nice when the grid goes down for a week after a CAT4 Cyclone. So, I hold a few lithium stocks, because there aren't many viable EV options without it, but I don't want a Tesla battery pack on my wall, unless the price comes down ridiculous amounts, if it won't last ten years.

[philosophical mode OFF]

hyphenated
Noobie
Posts: 5
Joined: Fri, 27 Dec 2019, 09:30
Real Name: Neil Carpenter

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by hyphenated »

Hi Antiscab

Yes - I considered just stacking six lead-acid gel batteries, but decided to play with some SuperCaps - I have four stacks of six Chinese 500F on route. The most cost-effective 'Branded' reputable higher-voltage units I can find are Tecate 16.2V PBL with active balancing, at three or four times the price - although they at least come with a datasheet and some specifications :-). If using a SuperCap solves the interworking, I'd be up for an upgrade, if only for longevity. Spending a few thousand on a 62V SuperCap module is not going to happen - my largess doesn't extend that far.

My intent (for the Chinese stacks) is to add some more protection and access for cell voltage checks, see if there any cells that aren't playing ball and perhaps swap out any that have gone negative (screw terminals) , then condition them with trickle and ramping up and down.

antiscab
Senior Member
Posts: 2601
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by antiscab »

hyphenated wrote:
Sat, 28 Dec 2019, 17:18
But the three main issues for me in a static environment in the subtropics are cyclical degradation, heat derating and fire risk. So, as a thought experiment, what is the expected percentage deterioration for the various lithium chemistries when operating daily charge/discharge and operating in a 25 to 35 degree high-humidity environment?
LiFePO4 would be down to around 65% of original capacity after 10 years, assuming no cooling, and the number of temperature excursions above 35 deg C aren't too long. 35 deg C isn't too bad, and high humidity isn't an issue.
LTO would have less capacity loss, though costs more.
NMC is cheaper, though degrades faster with cycling, and requires cooling due to higher capacity loss at temps above 30 deg C.
cooling can be something as simple as a cheap 1hp split system set to cool to 30 deg C, in an insulated room. probably couldn't hurt for LiFePO4 either, though you could get away with a higher temp limit to keep energy use down.
hyphenated wrote:
Sat, 28 Dec 2019, 17:18
I figure I need 20 kWh for day/night energy transfer, and a little more would always be nice when the grid goes down for a week after a CAT4 Cyclone. So, I hold a few lithium stocks, because there aren't many viable EV options without it, but I don't want a Tesla battery pack on my wall, unless the price comes down ridiculous amounts, if it won't last ten years.
Tesla wall should last 10 years - the chemisty used is more temp sensitive, however, it's actively cooled.

If you start with 30kwh you will still have 20kwh at the end of 10 years. you can still parallel more capacity later, allowing the original battery to last beyond 10 years (rather than doing a complete replacement)

I would still use LiFePO4 as it's more high temperature stable, and harder to turn into fireworks.
hyphenated wrote:
Sat, 28 Dec 2019, 17:44
I have four stacks of six Chinese 500F on route. The most cost-effective 'Branded' reputable higher-voltage units I can find are Tecate 16.2V PBL with active balancing, at three or four times the price - although they at least come with a datasheet and some specifications :-)
I bought 20 x 500F caps off aliexpress for testing, with a view to using 200 or so to replace the battery in a NHW10 prius.
https://www.aliexpress.com/item/4000239 ... b201603_53
None of the 20 I bought met specifications. all had less than 350F, most between 200F and 300F, and high internal resistance (I couldn't charge or discharge at more than around 10-15A). They also got rather hot.
The difference in capacity meant I couldn't use the full voltage.
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells

doggy
Groupie
Posts: 256
Joined: Tue, 24 Jul 2018, 09:59
Real Name: David Kerr
Location: Sydney
Contact:

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by doggy »

I also bought a bunch of SuperCaps last year and found the same sorts of issues. As well as the things mentioned by antiscab, I found the leakage too high and balancing them (both charge and discharge) was a pain and not very successful.

I am amazed that some US power companies seem to be successfully using them for Grid stabilisation.

On a different topic, I am rather interested in Flow Batteries. But not sure I'd buy one just yet. So I'll follow this thread with interest.

Cheers,
Dave

hyphenated
Noobie
Posts: 5
Joined: Fri, 27 Dec 2019, 09:30
Real Name: Neil Carpenter

Re: Interworking a Sunny Island with a Redflow Flow battery

Post by hyphenated »

Yeah - cheap supercaps come with a variety of net stories, including the theory that someone has been dumpster diving, and labelling the manufacturer rejects. Certainly the accepted wisdom is that the caps marked 500F are 350F or so. If this solution allows the Redflow and Sunny Island to get along, I'm up for moving to warrantied components.

Given that this is a short-duration energy store cum voltage placemarker, I'm hoping that lower capacitance and higher ESR are less important (famous last words) - the major concern I have is higher currents driving imbalance in the cells. I need to put a thermometer in the box to see if the caps get warm. The biggest excursion is on cold start (although I hope that the SI short-circuit protection will cope). If the highest capacitor string current we see is 5A that should suffice. :geek:

Post Reply