Agent Based Model of a simple dayahead electrictiy market

Written for a sumbisson to a PhD-Position programming challenge. Simulates a simple uniform dayahaed electricity market.

What should it do?

Simulation is called by MarketSim(fin='data/input_dummy.csv', fout='data/output_dummy.csv'):

• Takes as input a .csv-file includig the variables: time, demand, hydro_avil., wind_avail., solar_avail.;

• Outputs again a .csv-file containing 'time', 'system cost', 'marginal price' and the profit of all agents.

• Attention: if one of those variable names from the input-file isn't included, the code will not run! An exmaple is deilverd in data/input_dummy.csv.

Agents get initalized by MarketSim.add_agent(uid, tech, cap, var_cost, storage):

• uid: string, unique agent identifier

• tech: string, technology the agent uses. Possible choices are: gas, hydro, wind, solar, storage

• cap: int, installed capacity of this agent in MW

• var_cost: float, variable cost of generation in EUR/MWh. This holds the fuel cost (EUR/MWh_therm) for gas generators and the operational cost (EUR/MWh_el) for all other generators

• storage: int, optional, available storage capacity for storage agents in MWh

• It is not necessary to include all tech-types, if one (or more) are omitted a pseudo-agents gets initalized where all paramters are set to 0, so that it doesn't affect the outcome.

The Simulation gets executed by MarketSim.run(gas_efficiency, storage_efficiency, CO2_emission , CO2_price, bids_per_day):

• All offered bids and capacities are collected for every 15 minutes for the dayahead (i.a. 96 offers) and then get cleared step by step by merit-order market clearing.

• Bids of all agents are equal to their marginal costs, only the storage-agent bids the average marginal price (from the marginal prices of the day before) times the ratio of the newly purchased capacity and the maximum storage space + its marginal costs.

• Offered capacities for 'hydro', 'wind' and 'solar' are equal their initalized capacities times their individual availabilities from the .csv-input file.

• 'storage' always offers its available capacity/96 times storage_efficiency.

• 'gas' always offers its capacity times gas_efficiency and gets an additional penalty induced by CO2 emissions and CO2 price.

• Attention: before calling MarketSim.run() another time, execute MarketSim(fin='data/input_dummy.csv', fout='data/output_dummy.csv') again to reset the model.

Other functions included in class MarketSim:

• market_clearing: executes a merit-order market clearing. Input: demand and bids, Outputs: market price, bids and sold quantities.
• tie_break: If two or more agents shared the same bid, the function fairly divides the offered quantities between the agents within the tie break situation.