MILP Formulation for Dynamic Demand Response of Electrolyzers


METADATA ONLY
Loading...

Date

2022

Publication Type

Conference Paper

ETH Bibliography

yes

Citations

Altmetric
METADATA ONLY

Data

Rights / License

Abstract

Electrolyzers can reduce their electricity costs through demand response (DR) by adapting their production rate to time-varying market prices. Although the production rate can often be adapted rapidly, exploiting the full DR potential of an electrolyzer requires to consider slow temperature dynamics, leading to challenging mixed-integer dynamic optimization problems. In this contribution, we propose a dynamic ramping reformulation for real-time scheduling optimization of electrolyzers considering these slow temperature dynamics. Starting from a nonlinear dynamic model, the limits of the temperature gradient are derived to guarantee that the optimization result is feasible on the original model. The limits are then approximated conservatively by piece-wise affine functions leading to a mixed-integer linear program (MILP). Varying the number of piece-wise affine segments allows to explicitly balance model conservativeness against computational burden. We apply our reformulation to a validated alkaline electrolyzer model from literature. Our dynamic temperature ramping approach reduces production costs by 15.9 % compared to nominal operation. A quasi-steady-state optimization, which is restricted to production rates with steady-state temperatures in the allowed range, only leads to 12.8 % improvement. The proposed formulation achieves optimization runtimes below one minute, which is sufficiently fast for real-time scheduling.

Publication status

published

Book title

14th International Symposium on Process Systems Engineering

Volume

49

Pages / Article No.

391 - 396

Publisher

Elsevier

Event

14th International Symposium on Process Systems Engineering (PSE 2021+)

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Electrolysis; Demand response; Mixed-integer linear programming

Organisational unit

09696 - Bardow, André / Bardow, André check_circle

Notes

Funding

Related publications and datasets