Update dispatch for grid outages by brtietz · Pull Request #632 · NatLabRockies/ssc

brtietz · 2021-10-28T18:25:20Z

Brings the grid outage code into the mainline battery dispatch. Allows the resiliency runner and the outage dispatch to use the same code. Pairs with SAM pull request NatLabRockies/SAM#722, please pull dispatch_for_grid_outage branch to test.

Solves #631 and #590

Adds a second min SOC variable "batt_minimum_outage_SOC", so that the battery can have a different min SOC during nominal operation and an outage.
- Also prevents the SOC from going to zero during the default outage behavior, allowing the battery to black start the PV inverter in the morning
Add new output variables:
- batt_to_system_load - tracks the battery covering the inverter night time losses
- outage_losses_unmet - tracks unmet losses separately from the unmet critical load
Updates the DC resiliency runner to better account for efficiency updates to meet the critical load
Modifies the dispatch planning code such that the dispatch algorithms re-plan after an outage
Modifies the losses code in pvsamv1 so that the resiliency code can dispatch against AC losses
Updates the test values to account for all of the above

Recommended tests:

Ensure that grid usage is zero during outage periods
Ensure that the resiliency runner predicts when the battery will hit the minimum state of charge
Ensure the outage code works with pvsamv1, battwatts, generic battery and fuel cell batttery

…t and construct one. set up outage framework including outage manager in dispatch functions for automated btm and manual

… way, fix an efficiency declaration issue in battwatts, and a constructor issue in the powerflow code

…ns to the resiliency runner. Update the tests in cmod_battery_test to reflect the decreased reslience due to these restrictions. Make the restrictions more explicit in the resiliency code to save on iterations

…or_grid_outage

…d_outage

…veral bugfixes to both the outage dispatch and btm dispatch constraints, including appropriate test updates. custom dispatch is now further constrianed from discharging to the grid

…e critical load met

…use during outage

…or_grid_outage

…ariables to set up new tests, still need to update powerflow code

…osses when grid is down and battery is fully discharged

…d_outage

…M pr

…/NREL/ssc into dispatch_for_grid_outage

…om develop

…or_grid_outage

…rflow

…eet critical load. Add efficiency into interative loop for charging such that there isn't critical load unmet during charging steps

…or_grid_outage

…increase discharging or decrease charging to compensate

brtietz · 2021-10-28T21:44:56Z

It looks like I had to solve #614 to get one of the outage tests passing on Linux, so add that to the pile of issues this PR will fix.

…or_grid_outage

…s only updated once. Fix sign errors in DC charging that allow it to converge

…le and to/from grid

brtietz · 2021-11-01T21:24:55Z

The file I've been using to test this is attached here:
test_grid_outage_2.zip

sjanzou

I tested with your file posted and went through some other cases and found some interesting plots and the annoying desired duration in hours bug in the parametrics, kWh can be used with additional calculations

with some results that @n8blair might be interested in

@cpaulgilman may have an issue with the limits on the Battery page in the fuel cell and the dispatch page in the battery configuration (I do not see a way around that)

Hmmm.... Fuel cell is failing with an 8760 weather file and an 8760 load:

File for testing with failing fuel cell configuration
test_grid_outage_2_fuelcell.zip

sjanzou · 2021-11-02T10:17:22Z

        bool chargeOnlySystemExceedLoad,
-        bool dischargeOnlyLoadExceedSystem
+        bool dischargeOnlyLoadExceedSystem,
+        double SOC_min_outage


at some point we should stick with one naming convention and c++ instead of java structure. Just a comment and nothing specific to this pull request

sjanzou · 2021-11-02T10:19:39Z

    powerTempLoss_kW = tempLoss * m_numInverters * util::watt_to_kilowatt;
-    powerLossTotal_kW = powerDC_kW - powerAC_kW;
+    if (powerDC_kW < 0.0)
+        powerLossTotal_kW = fabs(powerDC_kW) - fabs(powerAC_kW); // DC connected grid charging


Are the charging and discharging losses all accounted for in the difference here?

This is the fix for #631

The more detailed loss accounting is in lib_battery_power_dispatch.cpp, scattered throughout the calculation functions. There's also code in cmod_battery that does updates for the actual AC losses in pvsamv1. Many of the changes in those three files are related to loss computation.

sjanzou · 2021-11-02T10:21:04Z


+    // Power converters and topology
+    batt_vars->batt_topology = ChargeController::AC_CONNECTED;
+    batt_vars->batt_ac_dc_efficiency = 96;


Is this really stating that changing and discharging efficiency is the same?

Sort of - this variable might be better named batt_ac_to_dc_efficiency. Since battwatts doesn't allow customization of these in the GUI, they are hardcoded to the detailed battery defaults.

Note that these lines have been moved to fix a big in the batt_power_charge/discharge kw (lines 209 and 210 in this PR), they haven't been changed.

… overwrites crit load in vartable

brtietz · 2021-11-02T23:14:20Z

The issue with fuel cell and generic system was that battstor was overwriting the values of crit_load before cmod_battery had a chance to read them. I've fixed that issue, but a few others have cropped up in the ssc tests. I'll fix those tomorrow, and it'll hopefully be ready for another round of review in the afternoon.

…mulations

…or_grid_outage

dguittet · 2021-11-04T13:12:16Z

The file I've been using to test this is attached here: test_grid_outage_2.zip

Thanks for adding. When I opened this, there were no critical loads specified or outage time periods (the 8760 was all 0), is that expected?

dguittet · 2021-11-04T14:05:04Z

I built on dispatch_for_grid_outage for both sam & ssc, but when I save the file and re-open it, the critical loads & grid outage timeseries inputs are back to the defaults. So something in the saving and reopening process isn't working for me. Is it just me?

I'm adding some high-level feedback in the comments before digging into the code:

It'll be nice to have these e-16 numbers rounded to 0. Probably by the ssc losses and/or powerflow code
I think the inverter is supposed to be off at the start of the year when there's no power production, but at the start of the second year, this behavior seems to revert to inverter nighttime losses again.
This label should be "power" not "energy"

…or_grid_outage

…ge value that said energy to power to match units. Set p_battery_ac to zero when it's less than tolerance in the powerflow code

brtietz · 2021-11-04T20:26:43Z

@dguittet Great catch on all of the above. Those should be fixed in the latest commit.

dguittet

Really well done on this large piece of work! And thanks for splitting up the grid outage work into separate PRs.

dguittet · 2021-11-05T21:05:37Z

-    if (pv_kwac > crit_load_kwac){
-        double remaining_kwdc = -(pv_kwac - crit_load_kwac) * m_batteryPower->singlePointEfficiencyACToDC;
-        remaining_kwdc = fmax(remaining_kwdc, max_charge_kwdc);
-        dispatch_kw(remaining_kwdc);


looks like dispatch_kw is no longer used. would it be ok to delete?

Yes, good catch!

dguittet · 2021-11-08T14:07:56Z

+        else if (h > 16) {
+            load_prediction.push_back(700);
+        }
+        else if (h > 18) {


this h>18 condition never comes on because anything > 18 is > 16 so will be set to 700

Right, thanks. Removing this doesn't change the test result, and these tests still show what I want them to show (ability to resume normal dispatch after an outage), so I've just removed this else if block.

brtietz · 2021-11-08T18:18:02Z

Merged in 9cf27a1

Brian Mirletz added 30 commits August 31, 2021 10:54

Merge branch 'more_flexible_powerflows' into dispatch_for_grid_outage

e8fb02b

add helper class to update powerflow variables for grid outage

3717b49

Merge branch 'more_flexible_powerflows' into dispatch_for_grid_outage

3dc5dad

more boilerplate code for the outage manager. bring it into dispatch_…

7d57dbb

…t and construct one. set up outage framework including outage manager in dispatch functions for automated btm and manual

implement full battery dispatch code for ac and dc outages. Along the…

d55b304

… way, fix an efficiency declaration issue in battwatts, and a constructor issue in the powerflow code

Merge branch 'develop' into dispatch_for_grid_outage

a440d85

including the full battery dispatch added additional power restrictio…

20d1acb

…ns to the resiliency runner. Update the tests in cmod_battery_test to reflect the decreased reslience due to these restrictions. Make the restrictions more explicit in the resiliency code to save on iterations

First test where a dispatch algorithm has to recover from an outage

d950fb4

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

9225243

…or_grid_outage

Merge branch 'connect_outage_vars_to_powerflow' into dispatch_for_gri…

2aac5cf

…d_outage

write tests for dispatch coming into and out of an outage. yeilded se…

71396b4

…veral bugfixes to both the outage dispatch and btm dispatch constraints, including appropriate test updates. custom dispatch is now further constrianed from discharging to the grid

fix code that determines whether or not to display outage outputs lik…

05f9554

…e critical load met

include crit load unmet in grid accounting to (mostly) zero out grid …

2f3b42a

…use during outage

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

6e4ae2c

…or_grid_outage

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

b6704b7

…or_grid_outage

partial implementation of power battery to system load. Have enough v…

6414526

…ariables to set up new tests, still need to update powerflow code

mostly solve AC nighttime losses, still need to do DC, and turn off l…

340b9f4

…osses when grid is down and battery is fully discharged

Merge branch 'connect_outage_vars_to_powerflow' into dispatch_for_gri…

be22596

…d_outage

tweak the outage code to work with the data array specified in the SA…

6c555e6

…M pr

add tolerance to grid charging check to allow for floating point issues

634d7d8

Merge branch 'connect_outage_vars_to_powerflow' of https://github.com…

bb7bbb5

…/NREL/ssc into dispatch_for_grid_outage

merge commit from connect_outage_vars_to_powerflow and latest code fr…

8904d9d

…om develop

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

819a6bb

…or_grid_outage

update reslience runner to deal with more complex losses code in powe…

c5a6468

…rflow

fix some merge issues that broke the outage code

a92160b

Add DC outage test. Increase power as appropriate so unit tests can m…

1a14c6f

…eet critical load. Add efficiency into interative loop for charging such that there isn't critical load unmet during charging steps

merge develop and new outputs into dispatch_for_grid_outage

0e9abec

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

d779c34

…or_grid_outage

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

c303b14

…or_grid_outage

pass expected ac loss percent through pvsamv1 into the battery code. …

f7615cb

…increase discharging or decrease charging to compensate

brtietz mentioned this pull request Oct 28, 2021

Add min outage soc NatLabRockies/SAM#721

Closed

This was linked to issues Oct 28, 2021

"Inverter total power loss" can be negative when DC connected batteries grid charge #631

Closed

Battery losses do not affect resilience runner #590

Closed

change pointer initialization to c++ 11 standard for linux

7da9f8e

brtietz mentioned this pull request Oct 28, 2021

Dispatch for grid outage and new min SOC NatLabRockies/SAM#722

Closed

Brian Mirletz added 2 commits October 28, 2021 14:38

convert all sorts in btm dispatch to stable sort

ba8d9fc

test tightening down tolerances for previously problematic test

d94a684

brtietz linked an issue Oct 28, 2021 that may be closed by this pull request

Price Signals Dispatch runs slightly differently on Windows and Linux #614

Open

Brian Mirletz added 3 commits November 1, 2021 10:25

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

a691e4a

…or_grid_outage

fix battery state issues in DC charging case - ensure battery state i…

c3532f2

…s only updated once. Fix sign errors in DC charging that allow it to converge

update inverter losses code to remove losses from unmet losses variab…

3cdf79d

…le and to/from grid

sjanzou requested changes Nov 2, 2021

View reviewed changes

Brian Mirletz added 2 commits November 2, 2021 16:59

fix bug in generic system and fuel cell where battery output variable…

38d4305

… overwrites crit load in vartable

update min soc and test results for battwatts

f775b59

Brian Mirletz added 2 commits November 3, 2021 09:22

fix generic system number of records calculations for non-lifetime si…

c8853df

…mulations

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

9600b62

…or_grid_outage

Brian Mirletz added 3 commits November 4, 2021 09:53

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

767afc5

…or_grid_outage

Merge branch 'develop' of https://github.com/NREL/ssc into dispatch_f…

b6583b2

…or_grid_outage

Fix issue where inverter losses were only removed in first year. Chan…

06f392b

…ge value that said energy to power to match units. Set p_battery_ac to zero when it's less than tolerance in the powerflow code

Update cycle count results now that micro cycles have been set to zero

c89a98c

dguittet approved these changes Nov 8, 2021

View reviewed changes

brtietz closed this Nov 8, 2021

brtietz deleted the dispatch_for_grid_outage branch December 7, 2021 16:50

Conversation

brtietz commented Oct 28, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

brtietz commented Oct 28, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

brtietz commented Nov 1, 2021

Uh oh!

sjanzou left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

brtietz commented Nov 2, 2021

Uh oh!

dguittet commented Nov 4, 2021

Uh oh!

dguittet commented Nov 4, 2021

Uh oh!

brtietz commented Nov 4, 2021

Uh oh!

dguittet left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

brtietz commented Nov 8, 2021

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

brtietz commented Oct 28, 2021 •

edited

Loading

brtietz commented Oct 28, 2021 •

edited

Loading