Design and Optimization of Flexible Hydrogen Systems in Refineries

Abstract

With the increasing demand for hydrogen resulting from fierce market competition and stringent environmental legislation, the hydrogen system has become an important component of a refinery. It is vital for the hydrogen system to be operated economically and safely under varying operating conditions. This calls for a systematic approach to the design and optimization of flexible hydrogen systems, which is the aim of this article. The hydrogen distribution network is designed at the minimum total annual cost subject to constraints on the flow rates and pressures of both existing and new equipment during the payback period. Varying hydrogen demands, different pipeline levels, and the possibility of hydrogen units being shut down are considered as operating conditions in the design optimization task, leading to the formulation and solution of a mixed-integer nonlinear programming (MINLP) problem. Using a linearization method, the MINLP formulation is approximated by a mixed-integer linear programming (MILP) problem, resulting in an acceptable quality and high efficiency. An industrial hydrogen system is taken as a case study. As shown in the case study, the proposed approach can handle high-dimensional and complex hydrogen system problems and gain significant economic improvements in comparison to an existing design.

Modeling and optimization of refinery hydrogen network – a new strategy to linearize power equation of new compressor

“…The equations to calculate the power of the new compressor(s) (Eqns and ) are nonlinear equations because the suction pressure (PI nc ), the discharge pressure PO nc , and the compressor inlet flow rate ( F in , nc ) exist as variables in these equations. All attempts to linearize these equations have so far been achieved by assuming suction and discharge pressures of the new compressor(s) as constant values (instead of existing as variables) . Here, these equations are linearized by taking the suction pressure (PI nc ) value equal to the lowest pressure in the entire network.…”

“…[27] Upon simultaneously utilizing Eqns (1)-(25), MINLP problem is obtained. The nonlinearity arises due to the presence of bilinear terms in Eqns (7), (17), (18), (20), and (23) as well as from the equations for the power of new compressor, that is, Eqns (13) and (14). The integer variables are introduced to determine if new equipments are indeed required by optimization.…”

“…All attempts to linearize these equations have so far been achieved by assuming suction and discharge pressures of the new compressor(s) as constant values (instead of existing as variables). [17][18][19]27] Here, these equations are linearized by taking the suction pressure (PI nc ) value equal to the lowest pressure in the entire network. Hence, suction pressure is set as a scalar variable in the optimization model.…”

“…The computation of this more complex problem was circumvented by the linearization of MINLP model via relaxing all of the bilinear and trilinear terms into linear constraints. Jiao et al [18] presented a flexible optimization approach to tackle the design and optimization of hydrogen network under varying operation conditions such as fluctuating hydrogen demand and during shutdowns of hydrogenconsuming units. MINLP problem was obtained in this work because bilinear and trilinear terms existed in the hydrogen balances.…”

Strategies of Merging Compressors for Capital Cost Reduction in Hydrogen Networks Synthesis

“…Ahmad et al (2010) presented an optimization model of multi-period hydrogen networks in which three phases were set in terms of the life of catalyst. Considering the varying hydrogen demands, different pipe levels and the possibility of hydrotreating units being shut down, Jiao et al (2013) presented an optimization of a flexible hydrogen network to minimize TAC. Jiao et al (2012) also proposed a multi-period scheduling model for hydrogen network to achieve the minimum operating costs.…”