Skip to content

Consistency of peak load at various aggregation levels #172

@gplssm

Description

@gplssm

In versioned DP dataset v.0.2.10 peak loads in load areas and LV grid districts deviate quite heavily.

Cumulative deviation in kW:

diff_residential      92549.608809
diff_retail          276115.733138
diff_industrial      -77962.741536
diff_agricultural    338403.227488

Model draft data of v0.3.0 shows smaller cumulative deviation

diff_residential      1.226731
diff_retail          -1.759502
diff_industrial      -5.175576
diff_agricultural    51.138129

But maximum deviation for agricultural and residential sector is large to be acceptable

Maximum deviation 'model_draft' in kW:

diff_residential     5.920596e+00
diff_retail          1.110223e-10
diff_industrial      5.551115e-11
diff_agricultural    5.408894e+01

Fix this in ego_dp_lv_consumption_peakload.sql

In addition, Ding0 does not consider load areas with a peak load smaller than 1 kW.
This raises two issues

  1. Data is inconsistent to LV grid district peak loads
  2. Data of of EHV/HV is inconsistent with MV/LV peak load data

"1." Could be solved in Ding0, but does not solve "2.". I don't have a proposal how to resolve this. Any ideas @ulfmueller @IlkaCu or @lukasol ? From my perspective its not an urgent issue, but it should be resolved at some point...

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions