Nonlinear model predictive control and H robust control for a post-combustion CO2 capture process

doi:10.1016/j.ijggc.2018.01.015

The key objective of this lab-scale study was to prove the technical feasibility of the membrane- and adsorption-enhanced water gas shift (WGS) reaction process that employs a carbon molecular sieve (CMS) membrane reactor (MR) followed by an adsorption reactor (AR) for pre-combustion carbon dioxide (CO2) capture, while demonstrating progress towards achievement of the overall fossil energy performance goals of 90% CO2 capture rate with 95% CO2 purity, at a cost of electricity of 30% less than the baseline capture approaches. Such a WGS system combining a MR and an AR in tandem can produce an ultra-pure hydrogen (H2) product (withoutmore » the need for using a post-processing step) continuously, until the adsorbent (hydrotalcite-based) in the AR unit is saturated for regeneration via a pressure swing adsorption (PSA) and/or temperature swing adsorption (TSA) operation. The project effort was led by the University of Southern California (USC) with assistance from the University of California - Los Angeles (UCLA) and Media and Process Technology, Inc. (M&PT). The project was divided into two Budget Periods (BP). During Budget Period 1 (BP1) of the project, the key focus was to: (i) design, construct and test the lab-scale experimental MR-AR system, (ii) select and characterize appropriate membranes, adsorbents and catalysts, and (iii) upgrade and experimentally validate an in-house mathematical model. The key focus during BP2 was to experimentally test the proposed novel process in the lab-scale apparatus using simulated fuel gas (gasifier off-gas), and to complete the initial technical and economic feasibility study. In each Budget Period, there were a number of milestones and success criteria, which were all met by their due dates. Specifically, during BP1, the team: (i) completed the construction of the lab-scale MR-AR experimental system; (ii) prepared and characterized CMS membranes at the anticipated process conditions; (iii) prepared and characterized adsorbents at the anticipated process conditions, and generated global rate expressions for the catalyst; (iv) began testing of the MR and AR individual subsystems; and (v) developed mathematical models and began validating their ability to fit the experimental data. During BP2, the team: (i) experimentally tested the proposed novel process in the lab-scale apparatus using simulated fuel gas (gasifier off-gas), and (ii) completed the initial technical and economic feasibility study. Additional details on some of these key technical accomplishments are provided below. The team prepared high-performance CMS membranes for use in the membrane reactor testing activities. These membranes displayed very high performance and met all the original project performance targets: H2 permeance (1 to 1.5 m3/m2/hr/bar, or 370.3 to 555.5 GPU) and a H2/CO selectivity >80 at the relevant temperature (up to 300oC) and pressure conditions (up to 25 bar). They have been shown to be stable in intermediate-term thermal stability testing (consistent with our past experience). Further, the team had success in developing very highly H2/CO2 selective membranes (H2/CO2 selectivity >>200) using our higher quality tubular substrates in conjunction with modification of the pyrolysis conditions. Hydrotalcite (HTC) adsorbent materials were also prepared and characterized under both atmospheric and high (up to 30 bar) pressure conditions at M&PT and at higher pressures at USC. These materials have shown large CO2 uptake capacities (>10 wt%) and working capacities under cyclic AR conditions (3 wt%, the targeted value in the project). They have also been stable during CO2 cycling in various atmospheres, including a >500-hour MR-AR run. A commercial sour-shift catalyst was utilized in the experiments. Global rate expressions based on the experimental data generated were formulated and used in the models developed for simulating the behavior of the lab-scale MR-AR system as well as in the generation of a preliminary process TEA. A key focus during the initial phase of the project was to modify an existing lab-scale test unit at USC to permit operation at higher pressures (up to 25 bar) and to incorporate a dedicated AR subsystem. This new lab-scale system consisted of: (i) the gas delivery system; (ii) the MR system containing the CMS membrane packed with WGS catalyst in its annulus; (iii) the AR subsystem, consisting of two adsorbent beds for continuous cycling between adsorption and desorption, and containing the adsorbent and catalyst with its appropriate valves and control hardware; (iv) overall system control hardware; and (v) the analysis section equipped with the appropriate analytical equipment. Graphite O-rings were employed as seals between the membrane and housing. GC-TCD and GC-FID were used in concert with MFM to study the performance of the MR subsystem by measuring the total molar flow rates and analyzing the permeate and feed/reject side gas compositions. Since the AR subsystem operated under transient conditions, mass analyzers were used to track its performance, together with GC analysis at preset intervals (for validation of the mass analyzer’s readings). Initial testing focused on the individual components (i.e., MR and AR) of the MR-AR lab-scale system. Key experimental parameters investigated for the MR subsystem included: (i) the membrane permeation characteristics; (ii) reactor temperature, feed and permeate pressures, contact time, catalyst weight to CO molar flow rate (W/FCO), and the permeate side (steam) purge rate. For the AR subsystem, key parameters studied included: (i) the catalyst and adsorbent weights; (ii) the operating envelope (i.e., temperature and pressure) of the adsorption/reaction cycle; and (iii) the desorption/regeneration mode. The dependent variables monitored included exit stream compositions and flow rates, and based on such data we quantified the CO conversion, H2 recovery and purity, CO2 recovery and purity, and degree of adsorbent regeneration. The experimental data were then used to validate a mathematical model developed as part of this project. Upon construction of the lab-scale system and validation of the performance of its individual subsystems, the integrated MR-AR system, which consisted of a MR followed by two ARs, was investigated. This integrated MR-AR system was experimentally evaluated for the WGS reaction in the context of the IGCC power generation application. The CMS membrane exhibited very robust and stable performance during the long-term run (over a >500 hr run of H2S exposure at 25 bar of pressure) and maintained high He/N2 (~126) and H2/CO (~100) selectivities over a total of 742-hour of H2S exposure during the MR-AR run. The combined MR-AR lab-scale system was tested during numerous multiple-cycle runs and displayed superior performance to that of a conventional PBR with high purities for the hydrogen product, which can be directly usable in a hydrogen turbine for power generation. During the MR-AR multiple-cycle run, the team continued to monitor MR subsystem performance with respect to MR CO conversion, H2 purity and recovery in order to verify both membrane and catalyst stability. The membrane and catalyst in the MR both displayed stable performance during the long-period run. To properly characterize and monitor the AR behavior, the team has defined the “AR effect time” in two different ways, and during the MR-AR multiple-cycle run, the team found that the “AR effect time” takes a few cycles to settle to its eventual “steady-state” value. The team also monitored the “pseudo-steady state” CO conversion in the AR prior to being switched into the regeneration mode by allowing for the sorption-enhanced WGS reaction time to be long enough so that the adsorbent becomes saturated with CO2. The team also found that it takes again a few cycles before the conversion settles down to its eventual steady state value. Further, the catalyst in the AR demonstrated very robust and stable performance during a continuous 18-cycle experiment under various regeneration treatment. Thus, one concludes that the membrane, catalyst and adsorbent are very robust and stable under the large concentration H2S, high-temperature and high-pressure IGCC-like environment during the long-period MR-AR multiple-cycle run. The aforementioned experimentally validated mathematical model was used to carry out a preliminary TEA of the MR-AR-based IGCC plant, which delivered the following key results: (i) 92% CO2 capture with a net power output of 593 MWe. This corresponds to a 9% increase in power output, while delivering greater CO2 capture, over the baseline IGCC plant which features a 90% CO2 capture with a net power output of 543 MWe; (ii) a COE of 113.1 $/MWh (86.3 $/MWh) corresponding to a $39.3/ton ($5.10/ton) CO2 capture cost, when Nitrogen is not sold (sold at $30/ton) and the MR membrane lifespan is considered to be 10 years. This represents a 16% (36%) reduction in the baseline IGCC w/CCS COE of 135.4 $/MWh and a 37.8% (91.9%) reduction in the baseline IGCC w/CCS CO2 capture cost of $63.2/ton; (iii) a MR-AR IGCC COE increase of 0.2% and 0.6% (from 86.3 $/MWh to 86.5 $/MWh and 86.8 $/MWh) if the MR membrane lifespan is reduced to 5 years and 2 years, respectively; (iv) a baseline IGCC w/CCS COE decrease of 16%, 17%, 36%, and over 100% (from 135.4 $/MWh to 113.1$/MWh, 112.2 $/MWh, 86.3 $/MWh, and -255.8 $/MWh) for N2 sale prices of $0/ton, $1/ton, $30/ton, and $414/ton, respectively.

This content will become publicly available on February 4, 2020

Title: Nonlinear model predictive control and H robust control for a post-combustion CO2 capture process

Citation Formats