Stochastic optimization-based sustainable retrofit of petrochemical
energy systems under multiple uncertainty
Abstract
We report a multi-objective stochastic mixed-integer non-linear
programming (MINLP) framework for sustainable retrofit and capability
expansion of traditional energy systems in petrochemical complexes.
Multiple uncertainties including energy demands, solar radiations and
wind speeds are considered in the optimization framework, which are
characterized by normal distributions of historical data or normal
distributions pre-defined with assumed mean values and standard
variations. A stochastic reduced order model sampling technique is
introduced to describe the uncertainties by a small number of scenarios
and their individual probabilities. The optimization framework further
accounts for system configuration selection and sizing of the candidate
energy conversion equipment, such as thermal storage units, gas
turbines, boilers, steam turbines, as well as their operating capacities
in each time period. A case study is investigated to demonstrate the
performance of the proposed strategy and the optimization results under
three modes (deterministic, stochastic and semi-stochastic programs) are
compared.