Weight and power optimization of steam bottoming cycle for offshore oil and gas installations
Introduction
Today's offshore oil and gas installations are mostly powered by simple cycle GTs (gas turbines). To counter the cost of CO2 emissions in Norway (taxes and quota), an alternative to a simple cycle configuration could be a combined cycle plant to increase the plant's efficiency and decrease the CO2 emitted per generated kWh. A steam bottoming cycle, as part of a combined cycle, needs to be simple, with low weight and volume, on an offshore oil and gas installation [1]. On a small scale, a few offshore installations have combined cycles installed [2]. A 2013 increase in the CO2 tax by the Norwegian parliament may make combined cycles more attractive for the future on the Norwegian continental shelf [3].
One of the keys to the implementation of combined cycles on offshore oil and gas installations is for the steam bottoming cycle to have a low weight-to-power ratio. For the remainder of this paper, the steam bottoming cycle will be referred to as the HRSC (heat recovery steam cycle). The HRSC consists of a HRSG (heat recovery steam generator), a steam turbine, a condenser, various pumps, a water treatment unit, and associated auxiliaries. While the design criteria for maximizing the HRSC power and efficiency are well-known [e.g., see Ref. [4]], those for minimizing the weight-to-power ratio are still unclear. Previous studies of off-shore HRSC installations are based on knowledge-based designs relying on previous experience, literature search, and experts' opinions as exemplified in Ref. [5]. The knowledge-based design methodology is described in Ref. [6].
Direct-search algorithms are widely used in engineering when the objective function is a black-box, e.g., a sequential flowsheet simulation code, or a solver of differential-algebraic equations. Black-box optimizers do not make use of derivative information (they are also called derivative-free methods) as the black-box function may be non-differentiable, discontinuous, not defined in some points of the feasible space, and affected by numerical noise. Well-known examples of such methods are the Simplex method [7], the Pattern Search Algorithm [8], the PSO (particle swarm optimizer) [9], and the several GAs (genetic algorithms) developed since the 1960s. A review and benchmarking of methods can be found in Ref. [10] for unconstrained and bound-constrained problems, and in Ref. [11] for nonlinearly constrained problems. Thanks to their robustness regarding numerical issues, such as numerical noise and discontinuities in the objective function, black-box methods have been successfully applied to several process engineering problems since the early 1970s, including steam cycles [12] integrated HRSCs [13], and steam generators [14].
More recently, engineering problems with different possible decision criteria (e.g., minimum weight or cost versus maximum power or efficiency), in which it is not easy to identify a single objective function, are often formulated and tackled as MO (multi-objective) optimization problems by means of evolutionary algorithms [15]. Instead, as for single-objective methods, such as the above mentioned direct-search methods, which return a single solution, evolutionary MO algorithms aim at determining the so-called Pareto frontier, i.e., a set of the most interesting solutions. Indeed, by definition of Pareto-optimality, no feasible solution gives better objective function values than a Pareto-optimal solution for at least one of the objective functions. For example, given two objective functions, f1 and f2 (e.g., power and weight), if the solution x is Pareto-optimal, no feasible solutions exist with the same value of f1 and a better objective function value of f2 and vice versa. Compared to the single solution returned by a single-objective optimizer, knowing the Pareto frontier is very useful in practice because: 1) it indicates not just one but a space of good solutions, allowing the designer to select the solution which best matches the installation constraints; 2) it graphically plots the trade-off between the different objectives (e.g., the gain of power which could be achieved by increasing the weight by a certain amount); and 3) it is possible to derive general criteria by analyzing the features of the Pareto-optimal solutions. As a consequence, MO algorithms have been extensively applied to the design of HRSCs in dual-pressure [16] and other combined cycles [17], Organic Rankine Cycles on offshore platforms [18] and for low grade waste heat [19], and novel energy systems [20].
The aims of this work were to find the minimum weight-to-power ratio for a heat recovery steam cycle designed for offshore oil and gas installations, and to determine the trade-off between weight and power. These aims were to be achieved while utilizing a commercial process simulator with detailed process models within an optimization framework. In more detail, a process model of a single pressure level combined cycle was coupled with MATLAB [21] and used as a ‘black-box’ function by specific optimization algorithms. The process model computed the combined cycle performance and weight for fixed HRSC design variables, which were set by the MATLAB optimizer. Within this framework, firstly PGS-COM, the direct-search algorithm proposed by Martelli et al. in Ref. [22] and described in detail in Ref. [11], was used to determine the solution with minimum weight-to-power ratio; then NSGA-II, the multi-objective optimizer described in Ref. [23], was applied to identify the Pareto frontier of solutions with maximum power and minimum weight.
Section snippets
Process description
The layout for the combined cycle was based on one GT (GE LM2500 + G4), one single-pressure OTSG (once-through heat recovery steam generator), one ST (steam turbine), and a deaerating condenser, as shown in Fig. 1. This setup is explained in more detail in Ref. [5] (layout c). The GT was equipped with dry low emission burners and VGVs (variable guide vanes). The use of VGVs for marine combined cycles is further described in Ref. [24]. Process model assumptions are listed in Table 1. The HRSG
Single-objective optimization
The objective function for the single-objective optimization was the HRSC weight-to-power ratio W/P.
One optimization required approximately 10,000 GT PRO simulations with a computational time of around 10 h on a 2.4 GHz Intel Core i7-2760QM CPU (Quad-core).
The results are displayed in Table 3. Compared to the knowledge-based design, the optimized solution has variables leading to a lower weight of the HRSG (weight-wise the dominant component) while allowing for a larger steam turbine and
Conclusions
Optimization of the design can open up new ways of thinking about the plant. A designer may have difficulty thinking outside the box, e.g., through being affected by existing designs and experts' opinion. This can be exemplified by the condenser pressure in this work. Existing designs in offshore oil and gas installations have a higher condenser pressure to keep the weight of the condenser lower. The knowledge-based design methodology, which is affected by previous designs, also selected a
Acknowledgments
Support from the Gas Technology Centre at NTNU-SINTEF helped achieve this collaborative effort. This publication forms a part of the EFFORT project, performed under the strategic Norwegian research program PETROMAKS. The authors acknowledge the partners Statoil, TOTAL E&P Norway, Shell Technology Norway, PETROBRAS, SINTEF Energy Research, NTNU-Norwegian University of Science and Technology, and the Research Council of Norway (203310/S60) for their support.
References (31)
- et al.
Design and off-design simulations of combined cycles for offshore oil and gas installations
Appl Therm Eng
(2013) - et al.
PGS-COM: a hybrid method for constrained non-smooth black-box optimization problems: brief review, novel algorithm and comparative evaluation
Comput Chem Eng
(2014) - et al.
Numerical optimization of heat recovery steam cycles: mathematical model, two-stage algorithm and applications
Comput Chem Eng
(2011) - et al.
Numerical optimization of steam cycles and steam generators designs for a coal to FT plant
Chem Eng Res Des
(2013) - et al.
Modeling and multi-objective exergy based optimization of a combined cycle power plant using a genetic algorithm
Energy Convers Manag
(2012) - et al.
Exergy, exergoeconomic and environmental analyses and evolutionary algorithm based multi-objective optimization of combined cycle power plants
Energy
(2011) - et al.
Multi-objective optimization of organic Rankine cycles for waste heat recovery: application in an offshore platform
Energy
(2013) - et al.
Multi-objective optimization of an organic Rankine cycle (ORC) for low grade waste heat recovery using evolutionary algorithm
Energy Convers Manag
(2013) - et al.
Methodology for the optimal thermo-economic, multi-objective design of thermochemical fuel production from biomass
Comput Chem Eng
(2009) - et al.
A novel hybrid direct search method for constrained non-smooth black-box problems
Comput Aided Chem Eng
(2013)