Sei sulla pagina 1di 10

international journal of hydrogen energy 35 (2010) 34993508

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Simulation of an hydrogen production steam reforming industrial plant for energetic performance prediction
A. Carrara, A. Perdichizzi, G. Barigozzi*
` degli Studi di Bergamo, Viale Marconi 5, 24044 Dalmine (BG), Italy Dipartimento di Ingegneria Industriale, Universita

article info
Article history: Received 7 October 2009 Received in revised form 23 December 2009 Accepted 24 December 2009 Available online 6 February 2010 Keywords: Gas separation Methane steam reforming Hydrogen

abstract
This paper presents the results of a theoretical investigation whose aim was the development of a simulation tool for performance prediction of a steam reforming hydrogen production plant, and particularly of its overall energetic efciency. A 1500 Nm3/h hydrogen production plant was simulated. Field data coming from an industrial plant were used for model validation in both design and off design operating conditions. To evaluate the plant performances in terms of energetic efciency, a particular attention was paid to the simulation of all plant auxiliaries consumptions. Nevertheless the large uncertainty in most of the eld data values, the model was able to capture all the relevant phenomena taking place in all the plant components, from reformer reactor up to CO2 sequestration unit, in the investigated plant capacity range (40100%). 2009 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

Alternative energy sources and power generation technologies are required to face the declining of fossil fuel stocks as well as the effects of carbon dioxide (CO2) emission on global warming. From these points of view, hydrogen (H2) is a very promising clean fuel, as no CO2 is produced by its combustion, available as fuel for distributed power systems [1], for example fuel cell systems [2]. Hydrogen is also an important raw material for the chemical and rening industries, e.g. for the production of ammonia and methanol. In the last years an increasing interest from the energy sector on hydrogen production techniques has been observed. Today most of hydrogen is produced from fossil fuel sources [3]. Clean production of hydrogen, for example by water electrolysis using renewable energy, seems not to be yet competitive with present-day renewable energy technologies. Hydrogen production from fossil fuel has as a consequence CO2 generation. Application of CO2 separation and

sequestration systems to hydrogen production process has often as a consequence a considerable reduction of process energetic efciency. For the near and medium term, the use of hydrogen as energy vector and fuel in distributed power plant systems needs to increase the energetic efciency of hydrogen production system from fossil fuels. About 50% of hydrogen production in the world today is based on methane steam reforming [4]. The methane steam reforming process is based on two main reactions: the reforming reaction CH4 H2 O CO2 3H2 DH298 206 kJ=mol (1)

and the water gas shift reaction (WGS): CO H2 O CO2 H2 DH298 41 kJ=mol (2)

In the last years many research activities have shown the potentiality of membrane technology [5]. Removing one or more of the products with a membrane would cause a shift in the reaction thermodynamic equilibrium, increasing the yield of CH4 and CO conversion [6].

* Corresponding author. 39 035 2052317; fax: 39 035 2052077. E-mail address: giovanna.barigozzi@unibg.it (G. Barigozzi). 0360-3199/$ see front matter 2009 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2009.12.156

3500

international journal of hydrogen energy 35 (2010) 34993508

Several research papers are available in the open literature considering the production of hydrogen by reforming (e.g. [710]), but most of them are focused on analyzing the reforming reactor only. Detailed simulations of methane steam reforming process are required to analyze possible plant modications (for example the inuence of membrane introduction [11,12] or changes in the feedstocks for hydrogen production [13]). The purpose of this work was to develop a simulation tool for design and off design performance prediction of a methane steam reforming industrial plant for hydrogen production to be used to critically evaluate the effects of dense membrane introduction on the hydrogen production process. The model was set up by using the commercial code Aspen Plus, which is suitable for chemical industrial processes. The reference plant is a 1500 Nm3/h hydrogen production plant located in the north of Italy. In the present paper the simulation tool is presented and its validation against eld data for variable plant capacity is demonstrated. This gave the opportunity to deeply analyze the off design behavior of the whole plant, in the meanwhile providing a useful tool for future investigation on the effects of the introduction of membranes on plant performances.

north of Italy with a nominal plant capacity of 1500 Nm3/h. The plant is structured in four main sections:     hydrogen production and compression; steam generation; combustion line; CO2 absorption and liquefaction.

2.1.

Hydrogen production and compression section

Following Fig. 1, the 21 bar and 350 K natural gas coming from the supply line through a compressor, is heated up by combustion products to about 650 K (CC-3) before entering the sulfur removal unit (S), making use of a ZnO bed. Sulfur is a powerful poison for nickel catalysts [14], whose maximum concentration in the feed gas to the reformer has to be reduced to less than 0.5 ppm. A small amount of hydrogen (0.03H2 to CH4 molar ratio) is added to natural gas to catalytically convert organic sulfur. In fact, the added hydrogen reacts with sulfur producing H2S, that is removed in the sulfur removal unit (S) where it reacts with the ZnO bed: H2 S ZnO ZnS H2 O (3)

2.

Reference plant description

The reference plant is the SIAD S.p.A. hydrogen production plant based on methane steam reforming (Fig. 1) located in the

Table 1 reports the mean natural gas composition delivered at plant location used in the present study. In the design condition reported in Fig. 1, corresponding to a 100% plant capacity, the hydrogenised natural gas mass ow rate is 422 kg/h.

Fig. 1 Reference plant lay-out (CC: combustion gas coolers (CC-1: steam boiler); DA: de-aerator; E: process gas coolers (E1: steam boiler); F: fans; P: pumps; R: Reformer; S: sulfur removal unit; SR: shift reactor; ST: stack; V1:steam drum; V2: water separation unit).

international journal of hydrogen energy 35 (2010) 34993508

3501

Table 1 Design methane composition. Element


CH4 C2H6 C3H8 n-C4H10 i-C4H10 n-C5H12 i-C5H12 C6H14 CO2 N2 He S (mg/Nm3)

2.2.

Steam generation section

Molar concentration %
96.34 1.79 0.41 0.08 0.06 0.01 0.01 0.01 0.09 1.19 0.01 5

Condensed syngas water mixed with feed water is pressurized to about 21 bar (P1 A/B), heated up (E4), de-aerated (DA) and then pumped (P2 A/B) and heated up again (E3) by the process gas before entering the LP drum (V1). From LP drum the feed water goes through the boiler for steam generation at 26 bar. Two boilers exist: the rst one (E1) makes use of the process gas exiting the reformer reactor, while the second one (CC-1) is located inside of the reformer convective section, recovering heat from the combustion gas coming from the radiant section. Steam is then superheated (E2) to about 570 K to be nally mixed with the compressed natural gas and directed to the reformer reactor.

After sulfur removal, methane is mixed with superheated steam with a design water to natural gas mass ow ratio r 3.296 kgw/kgCH4, and then heated up by combustion products to about 830 K (CC-2). The hot ow nally enters the reformer unit (R), consisting in a radiant and a convective section. The radiant section includes eight reaction tubes, with a Ni/CaO/Al2O3 catalyst, and two burners, making use of methane and purge gas as fuel. The process gas leaves the reformer at about 20 bar and at a temperature in the 1020 1130 K range. Its design composition is reported in Table 2. Before going to the shift reactor (SR), the process gas is cooled to about 540 K, providing heat to the boiler (E1) for steam generation and for superheated steam production (E2). The mean temperature (570670 K) adiabatic shift reactor makes use of a Cu based catalyst, with a ZnO and Al2O3 support to avoid sintering. The syngas then leaves the shift reactor at about 18 bar and 620 K. Syngas, H2 rich and CO poor, is then cooled down to about 310 K for water condensation (V2) in three heat exchangers (E3, E4, E5). The rst two use the water process as cooling ow, thus acting like an economizer providing pre-heated water to the boiler. The third instead makes use of well water. The dry syngas nally goes through the absorber unit for CO2 sequestration. The remaining syngas stream, CO2 poor, whose composition is reported in Table 2, enters the Pressure Swing Adsorption (PSA) unit, for hydrogen separation. In the PSA unit, thanks to the presence of molecular sieves and a cyclic process, 82% of hydrogen is separated form syngas. The remaining purge gas (Table 2) is used as fuel for the burners in the reformer radiant section, while the hydrogen ow is compressed from 16 bar to 200 bar.

2.3.

Combustion line

The purge gas is burned with air in the reformer radiant section, to generate the heat necessary for the reforming process. Unfortunately, the purge gas mass ow rate is not large enough to reach the right temperature in the reformer reactor, so natural gas is added. Purge gas combustion supplies up to 80% of process heat, while natural gas allows for bridgewall temperature control, that is the combustion gas temperature at the exit of the reformer radiant section. In design condition, the combustion products leave the reformer radiant section at about 1230 K. The gas then enters the convective section, where a boiler (CC-1) and three heat exchangers (CC-2, CC-3, CC-4) exist: one for natural gas heating before sulfur removal, one to heat the natural gas steam mixture before entering the reaction tubes and the latter, located at the end of the convective section, to pre-heat the combustion air. Combustion gases are nally released in the atmosphere. The presence of two fans assures the combustion products circulation through the reformer and the heat exchangers.

2.4.

CO2 absorption and liquefaction

After water separation, the syngas enters the absorption tower (Fig. 2); the two stage absorption process is based on the use of a mix of water and methyl-diethanolamine (MDEA). While the syngas, CO2 poor, goes through the PSA unit for H2 separation, CO2, water and MDEA ow is regenerated in a subatmospheric stripper tower. The pure CO2 ow then goes through a two stages intercooled compressor before liquefaction in a refrigerating thermodynamic cycle.

Table 2 Design gas compositions-plant capacity 100%. Molar concentration % Reformer exit
CH4 H2 CO CO2 H 2O N2 He 3.10 49.27 8.98 5.60 32.84 0.208 0.002

3.

The model

PSA inlet
4.91 89.82 2.534 2.06 0.34 0.327 0.003

PSA exit (purge gas)


18.64 61.37 9.63 7.81 1.30 1.24 0.01

The plant model was initially set up at design condition (100% plant capacity) using the commercial code Aspen Plus (version 2004.1). Aspen Plus software is a chemical engineering process modeling tool suitable for steady state simulation, design and performance monitoring. Aspen Plus simulation software was used for the mass and energy balance calculations. Plant model is structured, like the real plant, in four sections: hydrogen production and compression, water section, gas combustion line, and CO2 sequestration

3502

international journal of hydrogen energy 35 (2010) 34993508

Fig. 2 CO2 absorption process.

line. In the development of the model, eld data were used instead of design data for most components, every time design data were missed or resulted to be signicantly different from eld data. A particular care was devoted to the auxiliaries simulation, in order to predict the plant electrical consumption.

3.1.

Hydrogen production and compression section

The hydrogen production and compression section includes two ow inputs (natural gas and superheated steam) and four outputs: the pure hydrogen ow, the purge gas, the sequestered CO2 stream and the condensed water. Some are external

inputs and outputs, others are only ows internally exchanged with other plant subsections, like the superheated steam or the condensed water exchanged with the steam generation section. Fig. 3 schematizes the process gas path from steam to natural gas mixer up to PSA, including the CO2 separation unit (CS). Please note that natural gas (whose composition is given in Table 1) is rst pressurized in a 2 stage intercooled compressor (not shown in Fig. 3) up to 21 bar and 350 K before mixing with the superheated steam. A compressor isentropic efciency of 50% was assumed whatever plant capacity was concerned. To model both reforming and shift reactors, the RGibbs Aspen Plus reactor was used. The RGibbs reactor minimizes the Gibbs free energy to attain the equilibrium composition of the identied reforming or shift products (www.aspentech. com). Table 3 summarizes the reactions considered in the reformer simulation. Please note that the r value at design and off design conditions is always larger than 1.7, avoiding coke formation. For this reason carbon formation is ignored. According to gas composition analysis performed at the reformer reactor exit section, possible products (CH4, H2, CO, CO2, H2O, N2, He and Ar) are selected in such a way to obtain a complete reforming of all the hydrocarbons but CH4. Finally, to compute the hydrogen yield, the reformer pressure and temperature have to be specied. In the present simulation, the reformer pressure has been set to 19.7 bar, according to eld data, while the temperature has been dened using an iterative approach. Its value has been iteratively modied in order to obtain a design specied CH4 molar concentration at

Fig. 3 Hydrogen production section.

international journal of hydrogen energy 35 (2010) 34993508

3503

Table 3 Reactions taking place in the Reforming reactor.


CH4 H2O 4 CO 3H2 C2H6 2H2O 4 2CO 5H2 C3H8 3H2O 4 3CO 7H2 NC4H10 4H2O 4 4CO 9H2 IC4H10 4H2O 4 4CO 9H2 NC5H12 5H2O 4 5CO 11H2 IC5H12 5H2O 4 5CO 11H2 C6H14 6H2O 4 6CO 13H2 CO H2O 4 CO2 H2

Table 5 Design data for heat exchangers, PSA and H2 compressor. Natural gas supply temperature
Natural gas supply pressure CC-2 exit Temperature CC-3 exit Temperature E2 exit Temperature E5 exit Temperature PSA pressure PSA %H2 separation H2 compressor exit pressure H2 compressor efciency 1 Intercooler exit temperature 2 Intercooler exit temperature 3 Intercooler exit temperature

350
21 830 650 540 310 17 79 200 50 306 307 297

K
bar K K K K bar % bar % K K K

the reformer reactor exit section. The design specication for the CH4 molar concentration has been again obtained from measured data (4.0%). In the shift reactor of course the only possible reaction is the shift reaction (2). A major difference with respect to the reformer is that the shift reactor has a xed working pressure of 18.3 bar and it is an adiabatic reactor. The chemical equilibrium is anyways calculated dening the pressure value and the already mentioned possible products, with CH4 dened as an inert. Table 4 summarizes the main input data for both reformer and shift reactor simulations. All process gas coolers are simulated using classical heat exchangers, imposing the gas outlet temperature in the design condition and computing the recovered heat. Pressure losses are evenly distributed between the heat exchangers in such a way to obtain the design PSA operating pressure (17 bar). The PSA process is nally simply simulated using a splitter, xing the percentage of hydrogen separated from the dry syngas at about 79%. This value is slightly different from the design specication and was derived from eld data; moreover it was practically unaffected by changes in plant capacity. The hydrogen is then pressurized using a 3 stage intercooled compressor, from which it is delivered at 200 bar and 297 K. A 50% overall compressor efciency was used. Table 5 summarizes the main input parameters for the design of heat exchangers, PSA and hydrogen compressor unit.

aerator requirements. Fig. 4 shows all the components the water/steam ow encounters before mixing with the methane. It has to be observed that, in the real plant, two boilers operating in a parallel conguration do exist. In the present analysis the evaporative section was instead modeled using a serial conguration. In this way, not a complete evaporation takes place in the two boilers. In fact, the heat recovered from E1 heat exchanger, already computed in the previous section, is used to produce wet steam, that becomes saturated only at the exit of CC-1 heat exchanger. Finally, to correctly estimate the feed water pumps power consumption, manufacturer supplied performance curves were included into the model. All input parameters are summarized in Table 6.

3.3. 3.2. Steam generation section

Combustion line

The required water mass ow rate depends on the water to natural gas mass ow ratio r, the xed plant capacity and the natural gas mass ow rate entering the reformer reactor. Once the need of water for the reforming process is known, the system calculates the feed water mass ow, which depends on the condensed water and on the de-

Table 4 Input data and main assumption for reformer and shift reactor. Reformer
Inputs Assumptions Specications Possible products Property method Inert Restricted equilibrium

Shift Reactor

p 19.7 bar p 18.3 bar Adiabatic Restrict chemical equilibrium. Specify temperature approach or reaction. CH4, H2, CO, CO2, H2O, N2, He, Ar SRK IDEAL %CH4 1 See Table 3 Individual reaction: CO H2O CO2 H2

Fig. 5 shows the combustion products line while model input parameters are summarized in Table 7. Three input ows do exist (purge gas, natural gas and air) and one output ow (the combustion gas discharged into the ambient at the stack). The required natural gas mass ow is computed knowing the purge gas energy input and the reformer reactor energy requirement. The necessary air mass ow is then computed specifying the excess air 3 in the model input data. A 10% excess air applied to the 100% plant capacity design condition. The reformer radiant section is simulated using 2 units: an RSTOIC reactor, for combustion simulation, and a heat exchanger (CC-R) to simulate the heat exchange between combustion gas and reformer reactor. An RSTOIC unit is suitable to simulate a reactor when reaction kinetics are unknown or unimportant and when stoichiometry and the molar extent or conversion is known for each reaction. The reformer reactor (R) already introduced in the hydrogen production and compression section thus calculates the heat necessary for the reforming process while CC-R removes this heat from combustion gas, knowing the combustion gas exit temperature, i.e. the bridgewall

3504

international journal of hydrogen energy 35 (2010) 34993508

Fig. 4 Water section.

temperature. The air pre-heater CC-4 has been designed in such a way to heat the air up to 650 K. Finally, a constant fans overall efciency of 50% was used all over the considered operating range.

3.4.

CO2 absorption and liquefaction

The CO2 sequestration section is made of the ammine cycle (Fig. 2) and the CO2 liquefaction cycle. The ammine cycle has two input ows, the dry syngas and a mix of water and ammine, and two output ows, the CO2 poor syngas and the ammine CO2 rich ow. A mix of water, 50%wt, diethanolamine (DEA), 25%wt, and methyl-diethanolamine (MDEA), 25%wt was used [15]. The ammine mass ow delivered to the absorber tower was set so to have the design CO2 concentration in the syngas CO2 poor ow. A constant efciency of 60% was used in the water and ammine mixture circulation pump model. The CO2 rich ow is regenerated in the stripper tower. Before entering the stripper the CO2 rich ow is heated up by the hot regenerated ow. Both absorber and stripper towers are modeled with a RADFRAC unit. RADFRAC is a rigorous model for simulating all types of multistage vapor-liquid fractionation operations. Simulation model calculates reboiler duty so to have an exit ammine ow temperature of about 380 K. About 90% of CO2 is sequestrated; at a plant capacity of 1500 Nm3/h this results in 750 kg/h of pure CO2 generated. A two stages intercooled compressor (50% overall efciency) allows the pure CO2 ow to reach 17 bar and 295 K before liquefaction in a refrigerating thermodynamic cycle. In the

refrigeration cycle, the R507 working uid ow is compressed at 13 bar and 340 K, condensed and nally laminated before entering the evaporative section for CO2 liquefaction. Every detail of this cycle has been also included in the model, thus allowing to compute the electrical power consumption. In particular, the compression section has been simulated using two compressors, in parallel conguration, having an 83% isentropic efciency. Table 8 summarizes the input parameters to the CO2 absorption and liquefaction section model.

4.

The plant control strategy

Table 6 Water section input data.


Feed water temperature E4 exit temperature E3 exit temperature Evaporation temperature E2 exit temperature (superheated steam) 288 368 488 499 570 K K K K K

A high degree of automation characterizes the plant. The operator in fact only selects the desired plant capacity, while the control system automatically sets all the relevant parameters. In particular the plant capacity is allowed to vary within the range 30100%. The plant capacity specied, the control system modies the steam over carbon ratio r and the bridgewall temperature Tbw following the values reported in Fig. 6. It has to be observed that r has to be always maintained larger than 1.7 to avoid coke formation. To control the bridgewall temperature, the system in turns modies the methane and combustion air mass ow rates, knowing the required excess air 3 (Fig. 6). The combustion computation is performed in an iterative way: the model initially calculates the purge gas mass ow to combustion line and sets the air mass ow according to the prescribed excess air assuming no natural gas to the burner. The CC-R combustion gas exit temperature and the bridgewall temperature are then computed and the latter is compared with the prescribed value. In case a difference between computed and prescribed values does exist, a methane mass ow is added to the burner so to reach the correct bridgewall temperature. A modication in the plant capacity will also imply changes in pressure and CH4 molar concentration at the exit of reactor. Pressure and CH4 rate variations were then provided to the model (Fig. 6), allowing to calculate the reactor gas exit temperature and the reactor energy balance for variable plant capacity.

international journal of hydrogen energy 35 (2010) 34993508

3505

Fig. 5 Combustion gas line.

The laws of variation of all those parameters as a function of plant capacity were derived tting eld data with polynomial interpolation curves.

5.

Model validation

Figs. 35 report pressure, temperature and mass ow rate values of most signicant streams provided by the model at design condition. Table 9 reports the overall mass balance of the steam reforming plant at design condition. The mass balance of re-boiler of the stripper tower in the CO2 sequestration process is not considered. All the simulation outputs are in a reasonable agreement with eld data. In particular, one can observe that both reactor and shift reactor exit temperature values are correctly computed (1048 K and 617 K respectively), as well as the steam production (1429.5 kg/h). With a methane input of 433.7 kg/h, an hydrogen production of 135.9 kg/h is achieved as well as a CO2 production of 845.6 kg/h. The purge gas to the reactor (222.5 kg/h) is combined with a small methane mass ow (17 kg/h), giving rise to combustion gases exiting the stack at 570 K. The 659.2 kg/h condensed water ow coming from the water separation unit is combined with feed water to restore the required mass ow for steam production. The overall energy balance at design condition is reported in Fig. 7. Besides to reactor and burner, methane is also provided to the re-boiler of the stripper tower (75.2 kg/h) to regenerate the CO2 rich ow. Steam reforming section, CO2 sequestration and liquefaction section and H2 compression section all contribute in a similar way to the auxiliaries power consumption. The quite high loss level (about 3000 kW, roughly corresponding to 40% of energy input) can be ascribed

to losses in energy conversion and heat transfer processes, as well as in the heat rejection to the ambient through compressor intercoolers, in the liquefaction cycle and in reboiler stack. Moreover, it has to be underlined that, even if the energy value of CO2 is zero, its sequestration is very expensive. The same stands for the energy related to the hydrogen output: in the energy balance only its LHV is considered, but to be safely stored it needs to be compressed and, again, this requires an extra energy input. Once the model has been set up in design condition, off design runs were performed for validation purposes, comparing the code results against experimental data. Field data also including chemical composition of syngas in different plant locations as well as electric auxiliaries absorption were provided over a wide range of plant capacities, from 45% up to 100%. Before presenting the validation results, the reader has to be awarded that eld data here reported for comparison are mean values, as they have been obtained averaging, at constant plant capacity, the plant control panel data recorded over a one month time period. Deviations as large as 5% in some of the recorded parameters were observed. Moreover, the model input data for the off design computations have been also obtained by interpolating the corresponding averaged eld data. Finally, sensors precision is not always adequate for simulation purposes, even if it is so from a control point of view. All these aspects will inuence both

Table 8 CO2 sequestration and liquefaction input data.


Absorber top stage pressure Nr of absorption stages Stripper top stage pressure Nr of stripper stages Ammine ow exit temperature Circulating pumps efciency CO2 compressor exit pressure CO2 compressor exit Temperature CO2 compressor overall efciency Refrigeration cycle working uid Refrigeration cycle condensing pressure Refrigeration cycle compressor efciency 16.5 2 0.2488 2 380 60 17 295 50 R507 13 83 bar bar K % bar K % bar %

Table 7 Combustion line input data. 3


CC-R exit temperature (Bridgewall Temperature) Ambient temperature CC-4 air exit temperature

10
1230 288 650

%
K K K

3506

international journal of hydrogen energy 35 (2010) 34993508

Fig. 7 Plant energy balance at design condition.

Fig. 6 Off design specications.

experimental and numerical accuracy, and have to be considered when comparing simulation results against eld data. The model validation has been performed by comparing, at constant plant capacity, i.e. at constant hydrogen production, the measured and computed purge gas and methane mass ow rates to the burners. The purge gas mass ow rate will give information on the model capability of correctly control the methane steam reforming process, while the methane input to the burner will indicate its capability of control the combustion process. Both will inuence the overall plant efciency. Fig. 8 compares the purge gas mass ow rate values obtained from the off design simulations against eld data for variable plant capacities. An error as large as 25% in the purge gas mass ow rate prediction was obtained. This error progressively reduces approaching the design condition and it is mainly due to the measurement uncertainty. In fact, measurement uncertainties as large as 5% and 15% characterize respectively the dry syngas mass ow sensor, at the exit of the condense separation unit V2, and the dry syngas CO2 poor mass ow measurement, at the inlet of the PSA (see Fig. 3).

Fig. 9 compares the measured and computed sequestered CO2 mass ow rates for the considered plant capacities. In the simulation, a CO2 overproduction for small plant capacities can be observed, but in general the computed CO2 mass ow rate gives results in the range between mean and max values. This will imply that the error in the purge gas mass ow rate prediction is compatible with the measurement uncertainty. The large error in the purge gas mass ow rate inuences the natural gas mass ow rate going to the burners, as Fig. 10 shows. But the reported large errors do not exhibit a clear trend, due to very disperse eld data. However, the order of magnitude is correctly captured. The availability of CH4, CO and CO2 molar concentrations at different plant locations (reformer reactor exit section, PSA unit inlet section and in the purge gas) also allowed a model chemical validation. Fig. 11 shows the comparison between eld data and model results concerning the CO2 molar concentration at the reformer exit section. Only this concentration is relevant from a validation point of view. In fact, in the simulation the reformer reactor is modeled imposing the chemical equilibrium and changing the reactor temperature and pressure in such a way to reproduce the actual CH4 molar concentration at the reactor exit. The system then determines the heat requirement and the molar concentration of CO and CO2. As Fig. 11 clearly shows, the simulation capture pretty well the CO2 molar concentration, as differences lower than 1% take place over most of the

Table 9 Plant mass balance at design condition. Mass balance [kg/h] Input
Methane to Reactor Methane to burner Air Feed water 433.7 17 3536.3 770.3

Output
Combustion gas Hydrogen Sequestered CO2 3775.8 135.9 845.6

Fig. 8 Purge gas mass ow rate for variable plant capacity.

international journal of hydrogen energy 35 (2010) 34993508

3507

Fig. 9 Sequestered CO2 mass ow rate for variable plant capacity.

Fig. 11 Reformer exit CO2 molar concentration for variable plant capacity.

investigated domain. Larger differences only take place at the lowest plant capacity. Finally, the model prediction capability of auxiliaries (pumps, fans and compressors) electrical absorption was also validated against eld data. Unfortunately, beside the global plant auxiliaries power absorption, only the electrical power absorbed by the CO2 sequestration section is available from the plant control panel. Fig. 12 reports the comparison between simulation and eld data. A general good agreement between simulated and eld data characterizes both overall and CO2 section auxiliary consumption. The small difference existing between measured and computed global auxiliary power consumption is mainly related to the different operation strategy of some pumps: continuous in the model, discontinuous because coupled with storage tanks in the real plant. Finally, to allow for a comparison between different plant solutions, for example without and with the introduction of permeable membranes, an energetic efciency for the hydrogen production plant was introduced. It has been dened as the ratio between the energy of the produced hydrogen and the energy associated to all plant inputs: the methane mixed with steam introduced in the reformer, burned in the reformer radiant section and in the ammine regeneration process. Two denitions were used: a gross efciency, not considering the electrical auxiliaries power consumption (4) and a net efciency, also including it (5). To be

consistent with the other terms in the efciency formulation, the auxiliaries electrical power Pel,aux has been rst converted into its equivalent methane consumption, by means of a conversion efciency hconv of 43%, the average conversion efciency from natural gas to electricity suggested by the Italian Regulatory Authority for Electricity and Gas (www. autorita.energia.it) for 1025 MW sized power plants operating in the Italian scenario. mH2 LHVH2 hgross mCH4 Reformer mCH4 Burners mCH4 Stripper LHVCH4 (4)

hnet

mH2 LHVH2 mCH4 Reformer mCH4 Burners mCH4 Stripper LHVCH4 Pel;Aux =hconv (5)

Fig. 13 compares plant gross and net efciencies computed from the model against the same quantities derived from eld data for different plant capacities. First of all a very stable behavior of the hydrogen production plant can be observed, for variable plant capacity. An average gross and net plant energetic efciency of 63.5% and 52.1% respectively have been measured, with deviations of about 2.5% for both denitions. Moreover, the model prediction capability of the overall plant conversion efciency is quite good, as maximum differences of about 3.5% do exist between eld data and simulation results.

Fig. 10 CH4 mass ow rate to the burner for variable plant capacity.

Fig. 12 Electrical need of global plant and CO2 section for variable plant capacity.

3508

international journal of hydrogen energy 35 (2010) 34993508

Fig. 13 Gross and net plant electrical efciency for variable plant capacity.

6.

Conclusions

The present paper presents a simulation tool for design and off design performance prediction of a hydrogen production industrial plant based on methane steam reforming. This gave the opportunity to deeply analyze the off design behavior of the whole plant, in the meanwhile providing a useful tool for future investigation on the effects of the introduction of PdAg membranes on the energetic performances of the plant, which is being the topic of a following paper. To deeply investigate different plant solutions, a model was developed in AspenPlus environment. Many details have been included in the model, in order to simulate as close as possible all plant operational features, from thermodynamics properties, to chemical composition of main streams to electrical auxiliaries consumption. An energetic efciency has been also introduced to quantify the methane to hydrogen conversion process quality. The model, developed under design conditions, has been successfully validated over a wide range of off design operations, through a comparison against eld data. Some differences between real plant data and simulation results have been evidenced, but they have been mainly ascribed to a lack of accuracy in the eld instrumentation. Anyways, the model was able to correctly capture the trends of variation with plant capacity of all relevant parameters.

[3] Gambini M, Vellini M. Comparative analysis of H2/O2 cycle power plants based on different hydrogen production systems from fossil fuels. Int J Hydrogen Energy 2005;30: 593604. [4] Balat M. Potential importance of hydrogen as a future solution to environmental and transportation problems. Int J Hydrogen Energy 2008;33:401329. [5] Ferreira-Aparicio P, Rodriguez-Ramos I, Guerrero-Riuz A. On the applicability of membrane technology to the catalyzed dry reforming of methane. Appl Catal A 2002;237: 23952. [6] Tosti S, Bettinali L, Violante V. Rolled thin Pd and PdAg membranes for hydrogen separation and production. Int J Hydrogen Energy 2000;25:31925. [7] Dybkjaer I. Tubular reforming and autothermal reforming of natural gas an overview of available processes. Fuel Proc Tech 1995;42:85107. [8] Rakass S, Oudghiri-Hassani H, Rowntree P, Abatzoglou N. Steam reforming of methane over unsupported nickel catalysts. J Power Sources 2006;158:48596. [9] Seo Y-S, Shirley A, Kolaczkowski ST. Evaluation of thermodynamically favorable operating conditions for production of hydrogen in three different reforming technologies. J Power Sources 2002;108:21325. [10] Tong J, Matsumura Y. Pure hydrogen production by methane steam reforming with hydrogen-permeable membrane reactor. Catal Today 2006;111:14752. [11] Shirasaki Y, Tsuneki T, Ota Y, Yasuda I, Tachibana S, Nakajima H, et al. Development of membrane reformer system for highly efcient hydrogen production from natural gas. Int J Hydrogen Energy 2009;34:44827. [12] Iaquaniello G, Giacobbe F, Morico B, Cosenza S, Farace A. Membrane reforming in converting natural gas to hydrogen: production costs, part II. Int J Hydrogen Energy 2008;33: 6595601. cz E, Keiski RL. [13] Turpeinen E, Raudaskoski R, Pongra Thermodynamic analysis of conversion of alternative hydrocarbon-based feedstocks to hydrogen. Int J Hydrogen Energy 2008;33:663543. [14] Twigg MV. Catalyst handbook. London/PA: Manson; 1996. [15] Lombardi L. Life cycle assessment (LCA) and exergetic life cycle assessment (ELCA) of a semi-closed gas turbine cycle with CO2 chemical absorption. Energy Convers Manage 2001;42:10114.

Nomenclature
LHV: lower heating value, J/kg m: mass ow rate, kg/s p: pressure, bar Paux: auxiliaries electric power, kW r: water to methane mass ow ratio T: temperature, K Tbw: bridgewall temperature 3: excess air [%] h: energetic efciency [%]

references

` F. Decarbonized hydrogen and electricity [1] Consonni S, Vigano from natural gas. Int J Hydrogen Energy 2005;30:70118. [2] Perna A. Hydrogen from ethanol: theoretical optimization of a PEMFC system integrated with a steam reforming process. Int J Hydrogen Energy 2007;32:18119.

Potrebbero piacerti anche