[RE-wrenches] State of Charge Meter for Sol-Ark
Kent
kent at coveoregon.com
Fri Feb 9 15:18:40 PST 2024
William,
I think that the SOC determined by lithium BMSs come with similar
accuracy issues that occur with the Trimetric, FNDC, Magnum BMK,... -
small measurement errors integrated over a long time become big errors.
That's why Fortress (and every SOC meter) is saying that the battery
needs to reach 54.4 volts once a week for recalibration. At least with
lithium batteries the SOC meter doesn't need to account for the return
current dropping down as is required with lead-acid batteries. The
internal BMSs probably do a better job of estimating the SOC than these
external devices, but I do not expect them to be perfect. Same goes with
everyone's electric car, while we put a lot of faith in the displayed
SOC it probably isn't much better than ±5% and if it were off by 10% you
probably wouldn't know.
In regards to your graph showing a big voltage difference between two
55% SOC occurrences over a 12 hour time frame. I question your
assumption that the voltage should be the same. The data show different
voltages for the same SOC, it seems unlikely that the BMS measurement
drifted by enough to make that happen, so I think the data shown may
both be correct within reason. The voltage of lithium cells is highly
load dependent (probably somewhat temperature dependent too) so perhaps
the Fortress battery is actually close to right at both times.
Since using the SI SOC meter for starting the generator is problematic,
an external device for starting based on voltage is a good idea. If you
want a ready to go product to do that, I think the Morningstar relay
driver will work well. A little difficult to program or reprogram but
very reliable. One issue you will observe is that the generator won't
start at a consistent SOC as indicated by the battery or the SI.
Kent Osterberg
Blue Mountain Solar
On 2/9/2024 11:27 AM, William Miller via RE-wrenches wrote:
>
> Dave:
>
> I have to disagree with you here. To my knowledge every SOC system
> relies on coulomb counting and applying an efficiency value. That
> efficiency value is dependent on changing variables such as
> temperature, age of the batteries, charge rate, discharge rate-- to
> name a few. Every SOC system I have studied requires periodic
> recalibration. Maybe Discover is different but here is my experience
> with Fortress.
>
> Below is a chart of SOC versus Battery Voltage for a 4 Sunny Island
> system in a closed loop installation with 5 Fortress E-vaults:
>
> Inside the ellipse note that at 6:00AM the SOC is 55% and the battery
> voltage is about 50.8. At 4:00PM pm the same day SOC is again at 55%
> but the battery voltage is about 50.0 VDC. This is a drift in
> calibration of 0.8 VDC in 10 hours. In the context of LiFePo4 systems
> 0.8 VDC is a lot. Because this is closed loop, the drift was created
> in the BMS units, not the Sunny Island.
>
> Drake, to your question: The consequences depend on how much the
> system relies on SOC for operational mode decisions. I work almost
> exclusively with Outback and SMA SI and primarily in the off-grid
> segment. Since the SI is SOC-centric, the consequences experienced
> during the period charted above was that the system failed to start
> the generator and the system shut down.
>
> I am new to the use of Lithium batteries and don’t work a lot with SI
> so it took me two trips to figure out what was happening. The
> diagnosis was complicated by a failed SD card and a catastrophic
> generator failure (shorted windings). It took a few days to get a
> replacement generator and the interim the Fortress BMSs failed to
> recalibrate and this caused the incorrect correlation between VDC and SOC.
>
> In conversation with Fortress tech support I was advised to make sure
> the battery voltage gets to 54.4 at least once per week. This is the
> voltage at which the E-vaults recalibrate SOC. During the winter this
> is sometimes difficult to do. I can’t rely on the SI generator to
> auto-start and achieve 54.4 because it triggers on SOC values and if
> those values are wrong then the generator may shutoff prematurely,
> failing to reach 54.4. In this case the generator failed to start at
> all as a direct result of inaccurate SOC and the system crashed,
> causing an inconvenience to the client and to me.
>
> There are some settings on the SI that may force a full charge but I
> haven’t drilled down yet to see if this can be configured to ensure
> that 54.4 once per week. If anyone has experience with this and can
> chime in it may save me some time. I suspect all settings rely on SOC
> so there will be no built in solution.
>
> If the inverter here was an Outback and I was not relying on an FNDC
> to control generator start (which I never do), this problem would not
> happen. The Outback would react to battery voltage only, as monitored
> over three different time periods. This is a superior method for sure.
>
> Battery inverters are very smart these days, but sometimes not quite
> smart enough…
>
> Drake, thanks for asking.
>
> William
>
> PS: Brainstorming a solutions here:
>
> A: I am sure I could program an Arduino or Raspberry Pi to take over
> generator auto-start duties. I don’t prefer homemade solutions
> because of the time required to develop and test and I am not good at
> building interfaces.
>
> B: Maybe I could install a Mate3s and one FM60 charge controller. The
> charge controller would not connect to PV because this is an AC
> coupled system, but if were connected to the batteries it could
> monitor battery voltage and the internal aux relay could control the
> generator.
>
> Ideas, anyone?
>
> WM
>
> Miller Solar
>
> 17395 Oak Road, Atascadero, CA 93422
>
> 805-438-5600
>
> www.millersolar.com <http://www.millersolar.com/>
>
> CA Lic. 773985
>
> *From:*RE-wrenches [mailto:re-wrenches-bounces at lists.re-wrenches.org]
> *On Behalf Of *Dave Angelini Offgrid Solar via RE-wrenches
> *Sent:* Friday, February 9, 2024 9:29 AM
> *To:* RE-wrenches
> *Cc:* Dave Angelini Offgrid Solar
> *Subject:* Re: [RE-wrenches] State of Charge Meter for Sol-Ark
>
> Hi Drake,
>
> I do not think there is a lack of accuracy in a good closed loop LFP
> battery system. The steep curve of Lithium is just not accurate for
> voltage and the Soc measurement from a quality BMS is super accurate.
>
> All I use is the Discover AES and unlike others here, I do not have
> any issues over 4 years and 45 mostly Offgrid homes.
>
> The closed loop UL 9540 systems save alot of fuel in generator use,
> are simple, and have happy clients for me.
> *Dave Angelini Offgrid Solar*
> *"we go where powerlines don't"*
> *<http://members.sti.net/offgridsolar/>https://offgridsolar1.com/
> <https://offgridsolar1.com/><http://members.sti.net/offgridsolar/>*
> *e-mail offgridsolar at sti.net*
> *text 209 813 0060*
>
> On 2024-02-09 8:05 am, Drake Chamberlin via RE-wrenches wrote:
>
> How does the lack of accuracy in SOC detection affect the
> usefulness of closed loop systems?
>
> /Drake Chamberlin/
>
> /Athens Electric LLC/
>
> /Ohio Electrical Contractor’s License 44810/
>
> /NABCEP Certified PV Installation Professional/
>
> On 2024-02-05 11:36, William Miller via RE-wrenches wrote:
>
> Jason:
>
> I am careful about getting my clients too dependent on SOC
> readings. SOC is a calculated value based on changing
> variables and is notoriously inaccurate.
>
> Below is a screenshot of the Optics reporting for a client.
> The graph line that begins as the lower of the two is the SOC,
> the other is voltage. The SOC is out of calibration until
> about noon when it jumps from about 20% to about 80%. This
> does not mean the SOC changed by that amount, it means that it
> was just very wrong. Who knows when it is correct?
>
> In spite of repeated entreaties this client still reads the
> SOC and becomes concerned when it gets low-- even if the
> voltage level indicates the batteries are well charged. I
> have to deal with his misplaced anxiety.
>
> *Error! Filename not specified.**Error! Filename not
> specified.**Error! Filename not specified.*
>
> This problem appears to occur across all battery/inverter
> technology. For example, SMA touts their "coulomb counting"
> as more accurate than others but I have witnessed otherwise.
> You'd think that BMS units built by lithium manufacturers for
> their own products would be consistently accurate but even
> those BMS units need to recalibrate frequently, this according
> to the battery manufacturer's engineers.
>
> It would be nice to offer clients a simple, accurate method of
> ascertaining battery charge levels. SOC is not that method.
> I train my clients to watch voltage levels and to understand
> these values are elastic. If you can see trends in the
> battery voltage, so much the better. This is why I like the
> Outback Optics interface. This is also why a good AGS system
> examines battery voltage over time.
>
> I no longer install Outback FNDC units. Without them there is
> no SOC reading. I don't install Sunny Island systems—they are
> SOC centered and suffer for it.
>
> William
>
> Miller Solar
>
> 17395 Oak Road, Atascadero, CA 93422
>
> 805-438-5600
>
> www.millersolar.com <http://www.millersolar.com/>
>
> CA Lic. 773985
>
> *From:*RE-wrenches
> [mailto:re-wrenches-bounces at lists.re-wrenches.org] *On Behalf
> Of *Jason Szumlanski via RE-wrenches
> *Sent:* Sunday, February 4, 2024 7:30 AM
> *To:* RE-wrenches
> *Cc:* Jason Szumlanski
> *Subject:* [RE-wrenches] State of Charge Meter for Sol-Ark
>
> Are there any off the shelf solutions to view battery SOC via
> a wired meter mounted remotely on a property? I have a client
> with a simple voltage based meter for lead acid batteries that
> they are accustomed to using as a quick and approximate gauge
> of SOC. They want something similar for their new Sol-Ark with
> EG4 LL batteries.
>
> They will have smartphone app visibility, but they want
> something they can see inside the house without picking up a
> phone or going out to the inverter. Ideally the SOC will come
> from the inverter or the battery itself, not an external
> source (to avoid discrepancies).
>
> Jason Szumlanski
>
> Florida Solar Design Group
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.re-wrenches.org/pipermail/re-wrenches-re-wrenches.org/attachments/20240209/787330a6/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.jpg
Type: image/jpeg
Size: 153836 bytes
Desc: not available
URL: <http://lists.re-wrenches.org/pipermail/re-wrenches-re-wrenches.org/attachments/20240209/787330a6/attachment-0001.jpg>
More information about the RE-wrenches
mailing list