A cost assessment study for a large-scale water gas shift catalytic membrane reactor module in the presence of uncertainty

Abstract

A comprehensive economic performance evaluation framework for an actual large-scale watergas-shift Pd-based catalytic membrane reactor (CMR) module for hydrogen production is presented. Since a detailed assessment of the technical performance of the CMR built at WPI has been reported previously [1], the present research study focuses on an assessment of the module's economic performance characteristics, and thus, can be viewed as complementary. The proposed evaluation framework encompasses comprehensive baseline models for both Fixed Capital Investment (FCI) and Total Capital Investment (TCI) while various sources of uncertainty are identified whose effect on CMR's economic performance is explicitly taken into account using Monte Carlo techniques. As a result, insightful distribution profiles of FCI and TCI are derived rather than single-point value estimates and more realistic distributions of CMR economic performance outcomes are generated and statistically characterized. The latter could potentially inform development efforts of this new technology option for hydrogen production purposes.

Economic performance evaluation of process system design flexibility options under uncertainty: The case of hydrogen production plants with integrated membrane technology and CO2 capture

“…Fixed Capital Investment (FCI) encompassing both direct and indirect costs [31,32]. In particular, for the integration of the Pd/alloy-based CMR module into the above hydrogen production plant, cost components associated with direct costs involve purchased equipment (including WGS reactors, HTS catalysts, and Pd/Au alloy composite membrane tubes), installation, installed instrumentation and controls, installed piping, installed electrical units, buildings, process and auxiliary units, service facilities and yard improvements, as well as land.…”

“…Indirect costs include engineering and supervision pertinent costs, legal expenses, construction expense and contractor's fees, as well as contingency-relevant costs (for a more detailed description please see [9,29,31,32]).…”

“…avoided within the proposed context [23,40]. In addition, the above probabilistic distribution profiles can be also viewed as -risk-reward/opportunity‖ profiles that enable the quantification of -risks/rewards‖ through the -value at risk‖ and the -value at opportunity‖ at pre-specified levels, which are extensively used as financial evaluative tools/indicators in risk management and investment planning [9,29,32,40]. In the present study, the -value at risk‖ is denoted as P 5 , and it is characterized as a threshold-value at a probabilistic level of 5%, indicating that there is a 5% probability of generating a NPV lower than the P 5 threshold-value.…”

“…Triangular profiles are generally applied for cost estimation and risk assessments purposes with little knowledge about the model input outside the domain of approximate estimates related to its minimum, most likely, and maximum values[9,40,42,43]. Simple uniform or rectangular distribution profiles have a rectangular probability density function defined by two parameters, the model input's minimum and maximum values[29,32,40,44,45]. These profiles usually take into consideration reasonable ranges of values for the model input under consideration.…”

“…It should be pointed out that, in the absence of any reliable data at the commercial scale, the above simple uniform and triangular distribution forms considered in this study span the range of estimates reported in recent comprehensive studies/reports (cited in the list of bibliographical references). Finally, in the presence of available historical data, the aforementioned bootstrap distribution profiles commonly used in cost estimation and risk assessment studies were derived using standard re-sampling methods[9,32,40,46,47].The sequence of steps of the Monte Carlo simulation procedure followed in this study is depicted inFigure 3. The software package XLSim was used to conduct the Monte Carlo simulation studies with 10 4 runs each time and by taking into consideration all 40 uncertain model inputs simultaneously, as listed along with their corresponding probability distributions inTable 2.…”