"baseline" in ecm_prep_shapes.json changes when run with --tsv_metrics option

[gpavlak]

1) Prepare an ECM that includes "tsv_features": { "shape": { "custom_annual_savings": "<savings_shape_filename>.csv" } with the --tsv_metrics option.

python3 ecm_prep.py --alt_regions --site_energy --detail_brkout --tsv_metrics --sect_shapes --verbose

"tsv_metrics": ["1","2","2","1","1","1"]

mv ./supporting_data/ecm_prep_shapes.json ./supporting_data/ecm_prep_shapes_tsv.json
mv ./supporting_data/ecm_prep.json ./supporting_data/ecm_prep_tsv.json

2) Prepare same ECM without the --tsv_metrics option.

python3 ecm_prep.py --alt_regions --site_energy --detail_brkout  --sect_shapes --verbose
mv ./supporting_data/ecm_prep_shapes.json ./supporting_data/ecm_prep_shapes_no_tsv.json
mv ./supporting_data/ecm_prep.json ./supporting_data/ecm_prep_no_tsv.json

3) Compare the "baseline" and "efficient" data stored in ecm_prep_shapes_tsv.json and ecm_prep_shapes_no_tsv.json

4) Observation: "efficient" shapes are the same, but "baseline" shapes are different.

This may be related to parts of partition_microsegment where the effects of tsv_energy_eff and tsv_energy_base are factored out for the "efficient" case, but not the "baseline" case?

(around line 8150)

energy_total[yr] = \
                energy_total_sbmkt[yr] * diffuse_frac * tsv_energy_base
... 

self.sector_shapes[adopt_scheme][mskeys[1]][yr]["baseline"] = [
                    self.sector_shapes[adopt_scheme][mskeys[1]][yr][
                        "baseline"][x] + tsv_shapes["baseline"][x] *
                    energy_total[yr] for x in range(8760)]

(around line 8529)

 self.sector_shapes[adopt_scheme][mskeys[1]][yr][
                        "efficient"] = [self.sector_shapes[adopt_scheme][
                            mskeys[1]][yr]["efficient"][x] + ((
                                energy_tot_comp_meas +
                                energy_tot_uncomp_meas) / tsv_energy_eff) *
                            tsv_shapes["efficient"][x] + ((
                                energy_tot_comp_base +
                                energy_tot_uncomp_base) / tsv_energy_base) *
                            tsv_shapes["baseline"][x]
                        for x in range(8760)]

(Note division by tsv_energy_base and tsv_energy_eff in later chunk for the "efficient" case.)

example_files.zip

[jtlangevin]

Thanks @gpavlak for not only pointing out the issue but correctly identifying the solution! Fix to be committed shortly.

[gpavlak]

Excellent, glad to hear. Happy to make a small contribution to this already great work!