Susquehanna¶

The Susquehanna River, regulated by the Conowingo Dam, supports diverse needs, including hydroelectric power, water supply, and recreation. However, low-flow conditions create challenging trade-offs, forcing Conowingo to balance energy production with environmental and community water needs.

Originally, the simulation starts at (2021, 1, 1) and the decisions are made every 4 hours throughout a year resulting in 2190 time steps per episode (the whole simulation). It also has 6 following objectives:

Recreation (max), ref point: 0.0
Energy revenue (max), ref point: 0.0
Baltimore (max), ref point: 0.0
Atomic (max), ref point: 0.0
Chester (max), ref point: 0.0
Environment (min), ref point: -2190

Where max/main signifies whether an objective is to be maximised or minimised. The reference point is used for calculating hypervolume as the worst case scenario in terms of acquired rewards by the agent at the end of the simulation.

Observation space: Water level, month (2 dimensions)
Action space: Release per each reservoir (4 dimensions)

Here you can see a picture visualising Omo river structure: Alt text

Running¶

import mo_gymnasium
import morl4water.examples

water_management_system = mo_gymnasium.make('susquehanna-v0')

def run():
    #reset
    obs, info = water_management_system.reset()
    print(f'Initial Obs: {obs}')
    final_truncated = False
    final_terminated = False
    for t in range(10):
        if not final_terminated and not final_truncated:
            action = water_management_system.action_space.sample()
            print(f'Action for month: {t}: {action}')

            (
                        final_observation,
                        final_reward,
                        final_terminated,
                        final_truncated,
                        final_info
                    ) = water_management_system.step(action)
            # print(f'Final final_info: ', final_info)
            print(f'Observation: {final_observation}')
            print(f'Reward: {final_reward}')         
        else:
            break
    return final_observation
run()

Taken from: Muniak, K. (2024). RL4Water: Reinforcement Learning Environment for Water Management (Bachelor’s thesis). Delft University of Technology, Faculty of EEMCS.

Data & Data Sources¶

Data	Facility	Value	Source
max release	Conowingo (with 4 outflows)	41.302169, 464.16667, 54.748458, 85412	Salazar et al. (2024)
initial storage	Conowingo reservoir with Muddy Run Pump	2.6BCM	Salazar et al. (2024)
max capacity	Conowingo reservoir with Muddy Run Pump	7BCM	Salazar et al. (2024)
evaporation rate	Conowingo reservoir with Muddy Run Pump	Daily evaporation rates for Conowingo (see `examples/data/susquehanna/reservoir`)	Salazar et al. (2024)
storage to min max release	Conowingo reservoir with Muddy Run Pump	Tabular data (see `examples/data/susquehanna_river/reservoirs`)	Salazar et al. (2024)
storage to level to surface relationship	Conowingo reservoir with Muddy Run Pump	Tabular data for Conowingo	Salazar et al. (2024)
initial storage of the pump	Conowingo reservoir with Muddy Run Pump	1.9BCM	Salazar et al. (2024)
evaporation rates of the pump	Conowingo reservoir with Muddy Run Pump	Daily evaporation rates for Muddy Run (see `examples/data/susquehanna/reservoir`)	Salazar et al. (2024)
storage to level to surface relationship pump	Conowingo reservoir with Muddy Run Pump	Tabular data for Muddy Run Pump	Salazar et al. (2024)
pumping rules	Conowingo reservoir with Muddy Run Pump	Function `muddyrun_pumpturb` in main Susquehanna script	Salazar et al. (2024)
inflows to the pump	Conowingo reservoir with Muddy Run Pump	Daily inflow data to Muddy Run	Salazar et al. (2024)
efficiency	Conowingo Power Plant	79%	Salazar et al. (2024)
head start level	Conowingo Power Plant	0	Salazar et al. (2024)
n_turbines	Conowingo Power Plant	13	Salazar et al. (2024)
water demand	Atomice, Baltimore and Chester Irrigation Districts	Daily time series data with water demand	Salazar et al. (2024)
turbines	Conowingo Power Plant	Tabular data with min/max turbine flow per turbine	Salazar et al. (2024)
inflow	Later and Main Conowingo Inflow	Daily time series data of inflows	Salazar et al. (2024)