Optimizing the Power Purchase Cost Problem
Dr. Shruti Mantri
Email: shruti_mantri@isb.edu
ISB Institute of Data Science, Hyderabad, India
Vishali Reddy Nalla
Email: vishalireddy_nalla@isb.edu
ISB Institute of Data Science, Hyderabad, India
I. Introduction
Telangana has witnessed rising demand in power consumption with average power demand between 8,000 MW and 9,000 MW since March 2019. The demand has been rising consistently ever since the state government started supplying 24x7 free power to the agriculture sector from January 2018. In the year 2018–2019, power demand had touched 10,817 MW. Since the demand is set to increase further, the power generation plants need to generate power to meet demand. The per capita consumption of electricity in Telangana is about 1,507 units which is much higher than the national average of 1,222 units.
In terms of the power consumption, according to officials of Telangana Genco, on an average, everyday 164 million units of power was consumed in 2016–17, which rose to 179 million units in 2017–18. By November 2018, the consumption had increased to 189 MU and now it is recording over 200 million units a day.
Telangana State Electricity Regulatory Commission plays a major function of to regulate tariff in electricity market. All the parameters related to formulation of tariff like sales, power purchase, investments, O&M expenses needs to be regulated by making suitable regulations in alignment with the National Tariff policy and National Electricity Plan.
The state governments are struggling with high costs of power purchase due to an increase in price of coal, rail freight, rise in electricity demand, and an increase in short-term purchase costs, among others. According to data compiled by the Ministry of Power, Telangana has the highest power purchase cost during the period under review at ₹5.10 a unit while Karnataka comes second at ₹4.74 a unit whereas country average stands at ₹4.3. This motivates us to concentrate on the problem of reducing the power purchase costs to decrease the costs burden on government and offer better subsidy to deserving consumers.
The objective of this study is to minimize government subsidy, optimize power purchase costs thereby properly quantifying agricultural consumption estimate by applying different data science models. In the current study, researcher have designed and developed optimization model for optimizing power purchase cost.
Business Benefits of Optimizing Power Purchase Cost
Government purchases power in two ways: (i) Long term power purchase contracts (ii) Short term power purchase contracts.
(i) Long Term Power Purchase: The long-term agreements usually last for 20–25 years where both party’s buyer and seller agree on key contractual terms like quantity of power supplied, penalties, incentives, and termination date. This also means that buyer is bound to pay a fixed amount of money during the given tenure to the seller irrespective of number of power units supplied. Almost 90% of the electricity demand is met by Long term PPA’s only.
(ii) Short Term Power Purchase: When demand is not met by existing generating stations (with long term PPA’s), government looks for power purchasing from the open market to fill the demand gap of power supply. Therefore, the cost incurred will be the per unit power price at given point of time, so the cost incurred is also high in these cases. The short-term purchases usually account for 10% of power purchases. Due to presence of multiple sources of power generation, government need to look for cost minimization by optimizing purchase costs. Broadly, we can define costs and revenue for TSERC as follows:
(i) Revenue (+) — Bills collected from consumers
(ii) Costs (-) — Power Purchase Costs, Delivery costs, Equipment repair costs, Arrears from consumers, theft costs
In an ideal scenario, revenue generated should be great than costs. In the current study, the authors have concentrated on power purchase costs only.
II. Design System Components — Use Case Diagram/Activity Diagram/Sequence, collaboration diagram
TSERC needs to forecast’s demand of the power for every month and places order for the power supply with the power generation center as per the agreement. If the demand is not fulfilled from the existing power generation plant, TSERS places power either through short term agreement or open market bidding (Figure 2). The current working model of power purchase is illustrated in the activity diagram (Figure 2).
Flow-Chart of the Working System
III. Research Methodology
The goal of power purchase optimization is to minimum the cost of purchasing given quantity of electricity considering some constraints.
Where C = Cost per unit, P = No of units dispatched
The variables considered for optimization model are categorized as: (i) Cost, (ii) Generating Capacity and (iii) Demand.
Yellow — Variables in optimization function Green — Constraints
Data Source:
The data needed for the study was obtained from TSSPDCL and TSNPDCL.
Data Variables:
To design and develop an optimization model to minimize power purchase costs, following data variables were considered:
1) List of All generating stations registered with TSERC
2) Each generating plant capacity
3) Plant Load factor
4) Monthly forecasting demand
5) Losses in power transmission
6) Plant Type — conventional or non-conventional
7) If plant incurs fixed costs or not
The additional data needed was retrieved from the annual reports of power purchase costs and generating capacity.
IV. Model Designed and Developed
Optimization Model1: The model1 is built to optimize monthly power purchase costs.
· Parameters considered: All generating stations, total forecasted demand for month, costs, maximum and minimum units that can be dispatched (should be calculated from Generating Capacity & Load factor).
· Tools Used for Implementation: Microsoft Excel Solver
· Prototype: A prototype has been constructed to find the feasibility of the model. Due to unavailability of data, certain figures were assumed to run the model.
· Results and Findings: If the demand shoots up more than maximum capacity of all generators together then the model gives an exception or fails to find a solution.
Optimization Model2: The demand split for generators operating with Long Term PPA’s and Short-term power purchase/ on-spot power purchase
· To overcome the limitation mentioned in the previous model, the first phase of the model should find ideal percentages of demand that should be met by generators operating from Long Term PPA’s (L) and Short term or on-spot agreements(S)
Now the objective function is to minimize C, C = D1 * L + D2 * S
· Where D1 + D2 = Total Demand (D), L is average power purchase cost(unit/KWH) for Long Term PPA’s and S is average power purchase cost(unit/KWH) for Short Term or On-Spot agreements.
· Additional constraint, D1 <= maximum (generating capacities of all generators combined)
· This model will provide us the optimal percentage share of the demand that should be met by generators with long term PPA’s. This share might vary monthly due to presence of seasonal effects.
· Once, D1 is known; continue with Model 1 mentioned above.
V. Results and Findings:
1) In many scenarios power generators cannot be run at full capacity, so without considering more details on plant load factors for each generator, it is not feasible to practically consider number of units dispatched(P) to be a variable parameter
2) Energy Cost = Fixed + Variable + Incentive, where Variable cost again depends on units dispatched which cannot be practically controlled (given the constraints), so Cost is also not variable ©
VI. Challenges/Constraints Faced:
The following additional data will add more value to the model:
1) Insufficient Generating capacity data
2) Recommended maximum Plant Load factor
3) Hours of active generation from each plant
4) Demand forecasts
5) Season wise operational data of Conventional sources of energy
6) Transmission losses data
VII. Conclusion
Bulk electric power supply in India is mainly tied in long-term contracts. The DISCOMs who have the obligation to provide electricity to their consumers mainly rely on supplies from these long-term contracts. Nevertheless, to meet the short-term requirements of the market participants, short term trading plays an important role in the power market. However, recent trends show an increasing trend in per unit power purchase costs. This has raised concern and need for power purchase cost optimization. As we looked deeper into the variable parameters that could possibly reduce the power purchase costs, the regulations in place and the long-term contracts don’t offer much space for cost optimization in the current scenario. However, this model can be considered while renewing long term contracts or signing new contracts in future. The thumb rule is generators running with high Plant Load Factors can decrease per unit power costs, therefore this model can help eliminate low performing generators.
Acknowledgement
We would like to thank Telangana State Electricity Regulatory Commission for the sponsoring the project topic and thank Mr. Umakanta Panda,Commission Secretary FAC & Joint Director (Information Technology), Telangana State Electricity Regulatory Commission for his guidance and useful insights during the development of the project.
References
[1]https://www.uday.gov.in/images/Presentation%20by%20Tata%20Power.pdf
[3] http://downloads.hindawi.com/journals/mpe/2015/493845.pdf
[5] http://www.cercind.gov.in/2018/MMC/AR18.pdf
[6] http://www.cea.nic.in/reports/annual/annualreports/annual_report-2019.pdf
[8] Solving the Power Purchase Cost Optimization Problem with Improved DE Algorithm(https://www.researchgate.net/publication/319354354_Solving_the_Power_Purchase_Cost_Optimization_Problem_with_Improved_DE_Algorithm)
