Assessment of Model-Based Schemes for Accelerating Optimization by RAVEN

Technical Report · Thu Aug 30 04:00:00 UTC 2018

DOI:https://doi.org/10.2172/1490804· OSTI ID:1490804

Maronati, Giovanni ^[1]; Ponciroli, Roberto ^[1]; Vilim, Richard B. ^[1]

---

1. Argonne National Lab. (ANL), Argonne, IL (United States)

Nuclear-Renewable Hybrid Energy Systems (N-R HES) combine different energy technologies in a synergistic way with the objective of achieving a more economical energy production. In this report the particular hybrid system dealt with includes a PWR nuclear reactor, the associated Rankine-based energy conversion cycle (Balance of Plant, BOP), an Industrial Process (IP) for hydrogen production, a gas turbine (Second Energy Source, SES), and a battery (Energy Storage, ES). The RAVEN-based HYBRID framework is used to find the optimal installed capacity and the optimal economical dispatch of each component of the N-R HES. HYBRID solves the coupled capacity planning-economic and dispatch optimization problem with the objective of minimizing the Levelized Cost of Electricity (LCOE). In the optimization of the power dispatch strategy, the HYBRID framework takes into account the time-dependent dynamics of each subsystem by including a Dymola object-oriented model of the process dynamics. To ease the computational burden of the optimization process, an initial guess for the dispatch optimization problem is provided by a preconditioner. The solution guess is calculated by means of a least-marginal cost approach which applies the same criterion behind the screening curve method. Though this algorithm allows evaluating an economically adapted generation mix for a target load duration curve within a few seconds, it neglects the constraints due to the single component dynamics, and it is not able to adequately represent the ES. An improved treatment of the pre-conditioner problem was addressed by introducing a Monte Carlo based optimization algorithm. The new approach takes into account limits on power rates for the system subcomponents, which was not previously done. Additionally, this new Monte-Carlo based preconditioner addresses the problem with the ES component through improved modeling of that component. The new algorithm has been implemented in the HYBRID framework and tested to assess its performance. The results obtained with the new Monte Carlobased preconditioner were benchmarked against those obtained through the reference deterministic approach. The comparison shows that the Monte-Carlo based approach provides improved solution accuracy to the power dispatch optimization problem. Also in this report, the treatment of the penalty function on LCOE is discussed, and an improved penalty function for missed demand is suggested and tested.

• Research Organization: Argonne National Laboratory (ANL), Argonne, IL (United States)
• Sponsoring Organization: USDOE Office of Nuclear Energy (NE)
• DOE Contract Number: AC02-06CH11357
• OSTI ID: 1490804
• Report Number(s): ANL--18/31; 146826
• Country of Publication: United States
• Language: English