GCAM-USA Scenario Errors

[ecwood]

When I try to run the GCAM-USA reference scenario (this time, I did try to run it with the default dac_ssp2.xml and dac_USA_ssp2.xml, along with the carbon_tax_10_5.xml policy file), I get a lot of errors. I've attached the error log, up until the unreadable point (while I've posted a snippet of below). There are several GB of those types of unreadable errors: gcamexe_usa1_cropped.log

Error aEdding to demand in marketplace for: CO2, region: USA, value: rror adding to dnanema
nd in marketplacError adding to demae for: nd in marketplace for: SO2_1, region: USA, value: CO2, region: USA, vnanalue
: Error adding to denanmand in m
arketplace for: BC, region: Error adding to demanUSA, value: d in marketplace for: SO2_1, region: USA, value: nannan

Error adding to dEemand in marketplraror adding to demace for: BC, rndegion: USA, value:  in marketplace for: nanCO
, regionErro: r addinUgS to demand in marketAplace for: CO, region: USA, value: , value: nan
nan
ErrorE adding to demand in marketplace for: NH3, region: USA, value: rror adding to demand in marketplnana
ce for: NH3Error adding to dema, region: USA, valund in marketplace for: NMVOC, region: e:USA , value: nan
nan
Error adding tEro demand irn marketplace for: oNMVr adding OtCo, r demand in marketplace for: NOx, region: USA, valueegion: : USA, value: nan
nan
Error adErrdoir adding to demand in marketplace for: NOxng to ,dem reagionnd: U in marketplace for: SAOC, , region: USA, value: value: nan
nan
Error adding to demanError adddi in marketplace for: PM10, region: USAng, value:  to demand inna nmarketplace for: OC, region: USA, value:
in regcorrectlyrketion for sector . Sum is  did not normalShares for USAsector tError adding to depmanlace for: rade did notize traded natura normalil gze correctly. Sum isasNd coald in marke2Ot, region: pUS in reA0lace for: CcH4orrect * ely. Sum is , value: , region: x p( USA, value: 0 ) nan in region USAg0 did not normalion nan
USAi diError adding to
.0d not nor * expError adddeimand i
( malize correctl0y. Su )Errng to su * pplor  settinexp(ze correctly. Sum is n marketplacye in mm is  fog pricr: e in marketparketplace for: coal, region: USA, value: 0N2O,  *0 e 0 xp(na)  .lace for: refined liqui0ds
n
 regionEr: ror setting pri ) * e.xp(
 enduse0, value: E ) .rror adding to de
nance in marketplace f.or:
Error setting price inmand inErr ma marketplaceUSA, value: r for:
or setting price in marError setting price in rkemarkeefined liquids industriatplace ftptraded natural galace for: or:CO2 s, region: UStr, value: aded coalAnanketplace for:, value: nan lnantraded oil

, value: , value:
, value: nan
 ShA did nUSAare0s for sector * exp(, v  Errotraded oiot normaliz0e correctly. Sualue: m is  ) nan
rl a in region dding to demanError0 d in marUSA * exp(.
0adding  ) to deman did not normalize correctlketplace fd in ma.y. S
rketplaoce for: rBC: Error CO2adum is , region: USA, value: 0, regionann:
dEirngrUSA to demanod in marketplace for:  * exp( Error adding to demr addin, value: 0natural gnan
and  in marketplace for: ) CO.
, region: USAg toa demand in marketplace for: , vcoal, region: sU,S rA, value: egion: alnan
USA, ErError adding to deue: ror adding to demand in marketplavalue: nan
cnError adding to supply in manarketplmand in marketplace e
ace forError adding to de fofror: Error adding to supply in marketplSO2_1: ,tradeda ce: for:  cNH3, region:  USAregion: USAmand in mo, value: arketplace for:trad, nvaed natural gas,  region: lana
ue: lError ananddi
nUSA,, value: crudg  regto deEnane o

[ecwood]

When I installed GCAM on a fresh instance and ran the GCAM-USA Reference scenario, this was the output log: setup_gcam.log

It was a success.

[ecwood]

I copied the configuration_usa.xml file and changed it like this:

...
                <Value name = "transport_emiss_USA">../input/gcamdata/xml/transport_emissions_USA.xml</Value>
                <Value name = "prc_usa">../input/gcamdata/xml/ind_urb_processing_sectors_USA.xml</Value>
                <Value name = "process_emiss_USA">../input/gcamdata/xml/ind_urb_proc_emissions_USA.xml</Value>
                <Value name = "refining_emiss_USA">../input/gcamdata/xml/refinery_emissions_USA.xml</Value>

                <Value name = "solver">../input/solution/cal_broyden_config.xml</Value>

+               <Value name = "dac">../input/gcamdata/xml/dac_ssp2.xml</Value>
+               <Value name = "dac_usa">../input/gcamdata/xml/dac_USA_ssp2.xml</Value>

        </ScenarioComponents>
        <Strings>
                <Value name="scenarioName">GCAM-USA_Reference</Value>
                <Value name="debug-region">CA</Value>
                <Value name="MAGICC-input-dir">../input/magicc/inputs</Value>
                <Value name="MAGICC-output-dir">../output</Value>
                <Value name="AbatedGasForCostCurves">CO2</Value>
        </Strings>
        <Bools>
                <Value name="CalibrationActive">1</Value>
                <Value name="BatchMode">0</Value>
                <Value name="find-path">0</Value>
...

Period 5 of the model failed. Here's the output log: gcamexe_usa1.log

[ecwood]

I received this message from Jay Fuhrman:

Hi Erica, Celina,

Nice meeting you this morning. I spoke to my colleague at PNNL who led the development of DAC in GCAM-USA and he let me know that you shouldn’t have to worry about the errors for adding demand to marketplace in 2015. This is a known issue, but won’t affect the model results.

Addressing the convergence failures in the 2020 model period should be fixed by increasing the max iteration count in the “cal_broyden_config.xml” file upwards to perhaps 5000 or 7500.

Let me know if you have any more questions, this sounds like an exciting project!

Jay

This suggested change is addressed by f1d2e7f.