Drycooling

Continue improving the currently open cooling-tower vapour-recovery workbook. Inspect and reuse its existing cells, variables, weather data, assumptions, and results. Do not rebuild the workbook or discard existing analysis.

Add a dry-cooling comparison using the same IT load, heat-rejection load, climate, and operating hours as the existing wet-tower baseline.

Model two clearly labelled dry configurations:

1. A conventional air-cooled chiller, with COP varying with ambient temperature and temperature lift.
2. A warm-water dry cooler using fans and pumps without a compressor whenever ambient dry-bulb temperature plus the cooler approach permits it. Use mechanical assistance only during hours when the required loop temperature cannot otherwise be maintained.

Report fan-only hours, mechanically assisted hours, annual cooling electricity, total facility electricity, PUE, and percentage energy increase relative to the wet baseline. Assume zero direct evaporative cooling water for the fully dry cases unless an optional adiabatic-assist mode is explicitly enabled.

Use the Phoenix field comparison—PUE 1.84 for air-cooled chillers versus 1.63 for water-cooled and evaporative cooling, approximately 12.9% higher facility energy—only as a reasonableness check, not as a calibration target:
https://doi.org/10.1016/j.resconrec.2022.106194

Before comparing results, verify what the workbook’s existing 6.6% increase measures. Report it both as:

- percentage of total facility electricity;
- percentage of IT electricity;
- change in PUE;
- annual MWh.

Do not compare the 6.6% with the dry-cooling result until their denominators match.

Add upstream electricity-generation water consumption. Use operational water consumption, not water withdrawal.

Define:

- `E_facility = PUE * E_IT`
- `W_upstream = E_facility * grid_water_intensity`
- `WUE_site = W_site / E_IT`
- `WUE_source = (W_site + W_upstream) / E_IT`
- `incremental_upstream = (E_scenario - E_wet_baseline) * grid_water_intensity`
- `net_source_water_saved = onsite_water_saved - incremental_upstream`
- `fraction_shifted_upstream = incremental_upstream / onsite_water_saved`

Only the additional electricity relative to the wet baseline counts as water shifted upstream. The wet baseline already has its own upstream electricity-water footprint.

Make `grid_water_intensity` configurable. Use 4.35 L/kWh as a clearly labelled 2023 U.S. reference value, not a universal value. It includes reservoir evaporation allocated to hydropower, so expose that accounting choice and include these operational-consumption sensitivity cases:

- U.S. supplied-electricity reference: 4.35 L/kWh
- wet-cooled natural-gas combined cycle: 0.776 L/kWh
- wet-cooled coal: 2.60 L/kWh
- wet-cooled nuclear: 2.54 L/kWh
- utility solar PV: 0.0038 L/kWh
- wind: approximately zero
- custom local-grid value

Sources:

- DOE/LBNL U.S. data-centre report:
  https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report.pdf
- NREL generation-technology factors:
  https://openei.org/datasets/files/969/pub/1748-9326_7_4_045802.pdf

Produce:

1. A comparison table for the wet baseline, mine-water vapour recovery, conventional air-cooled chiller, and warm-water dry cooler.
2. Stacked bars separating onsite and upstream water consumption.
3. A water-versus-energy trade-off plot.
4. A sensitivity plot showing how the fraction shifted upstream changes with grid water intensity.
5. A concise conclusion stating whether each dry option uses more energy than the existing 6.6% result and how much of its onsite water saving is offset upstream.

Keep visible output concise. Save detailed hourly calculations to files and display only summaries and plots. Run all cells after editing and correct any errors.