Revista Mexicana de Ingeniería Química

CONTENIDO

Volumen 8, número 3, 2009 / Volume 8, number 3, 2009

213 Derivation and application of the Stefan-Maxwell equations

(Desarrollo y aplicación de las ecuaciones de Stefan-Maxwell)

Stephen Whitaker

Biotecnología / Biotechnology

245 Modelado de la biodegradación en biorreactores de lodos de hidrocarburos totales del petróleo

intemperizados en suelos y sedimentos

(Biodegradation modeling of sludge bioreactors of total petroleum hydrocarbons weathering in soil

and sediments)

S.A. Medina-Moreno, S. Huerta-Ochoa, C.A. Lucho-Constantino, L. Aguilera-Vázquez, A. Jiménez-

González y M. Gutiérrez-Rojas

259 Crecimiento, sobrevivencia y adaptación de Bifidobacterium infantis a condiciones ácidas

(Growth, survival and adaptation of Bifidobacterium infantis to acidic conditions)

L. Mayorga-Reyes, P. Bustamante-Camilo, A. Gutiérrez-Nava, E. Barranco-Florido y A. Azaola-

Espinosa

265 Statistical approach to optimization of ethanol fermentation by Saccharomyces cerevisiae in the

presence of Valfor® zeolite NaA

(Optimización estadística de la fermentación etanólica de Saccharomyces cerevisiae en presencia de

zeolita Valfor® zeolite NaA)

G. Inei-Shizukawa, H. A. Velasco-Bedrán, G. F. Gutiérrez-López and H. Hernández-Sánchez

Ingeniería de procesos / Process engineering

271 Localización de una planta industrial: Revisión crítica y adecuación de los criterios empleados en

esta decisión

(Plant site selection: Critical review and adequation criteria used in this decision)

J.R. Medina, R.L. Romero y G.A. Pérez

Revista Mexicanade Ingenierıa Quımica

1

Academia Mexicana de Investigacion y Docencia en Ingenierıa Quımica, A.C.

Volumen 14, Numero 2, Agosto 2015

ISSN 1665-2738

1

Vol. 14, No. 2 (2015) 543-552

ANALYSIS AND CONTROL OF A DISTRIBUTED PARAMETER REACTOR FORPYROLYSIS OF WOOD

ANALISIS Y CONTROL DE UN REACTOR DE PARAMETROS DISTRIBUIDOS ENEL PROCESO DE PIROLISIS

A. Morales-Diaz∗, A.D. Vazquez-Sandoval1 and S. Carlos-Hernandez2

lCentro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional, Av. Industria Metalurgica No.1062 Parque Industrial Ramos Arizpe Ramos Arizpe, Coah. C.P. 25900, Mexico, Grupo de Robotica y

Manufactura Avanzada.2Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional, Av. Industria Metalurgica

No. 1062 Parque industrial Ramos Arizpe Ramos Arizpe, Coah. C.P. 25900, Mexico, Laboratorio de Revaloracionde Residuos, Sustentabilidad de los Recursos Naturales y Energıa.

Received May 26, 2014; Accepted May 11, 2015

AbstractThis document introduces a methodology for the design of a Proportional Integral + Sliding Mode Control (PI+SM) andits application to a biomass pyrolysis process. The objective of the proposed controller is to promote the transformationof wood into gaseous products, which have high energy value. First, a distributed parameter model of a pyrolysis reactor(from specialized literature) is analyzed and modified in order to be implemented in Matlab. After that, a based sliding modecontroller is developed considering the reactor temperature as the controlled variable and the applied heat as the controlinput. Different scenarios are considered and the performance of the designed controller is compared with a classical PIcontroller and the process open loop behavior via numerical simulation. The results show that the proposed control strategyenhances the gaseous components in the process output.Keywords: pyrolysis, sliding mode control, alternative energy

ResumenEn este documento se presenta una metodologıa para el diseno de un control Proporcional Integral + control por Modosdeslizantes (PI+SM), ası como su aplicacion en un proceso de pirolisis de biomasa. El objetivo del controlador propuesto espromover la transformacion de madera en componentes gaseosos, los cuales cuentan con alto valor energetico. En primerlugar se analiza un modelo de parametros distribuidos planteado en literatura especializada, que corresponde a un reactorde pirolisis; este modelo es modificado e implementado en Matlab. Despues, se disena un controlador basado en modosdeslizantes considerando a la temperatura del reactor que se obtiene en cada iteracion discreta como la variable controladay el calor aplicado como entrada de control. Se toman en cuenta diferentes escenarios para evaluar el comportamiento delcontrolador y para comparar su desempeno via simulacion numerica, con la operacion en lazo abierto y con un controladorPI clasico. Los resultados muestran que la estrategia propuesta mejora la produccion de componentes gaseosos en la salidadel proceso.Palabras clave: pirolisis, control de modos deslizantes, energıa alternativa.

1 Introduction

The pollution global warnings and the hypothesisof the fossil fuels end have led the scientificcommunity to search for reliable alternatives. Oneof them is biomass since it is a renewable source

and it can be transformed in several productswith high energy value, see Vicencio-de la Rosa,(2015), Morales-Diaz et al. (2010) and Sacramento-Rivero et al. (2010). In this paper, biomassis understood as the residue or waste of a livingorganism which could be used for energy production.

∗Corresponding author. E-mail: america.morales@cinvestav.edu.mxTel. +52 (844) 4389600, Fax +52 (844) 4389610

Publicado por la Academia Mexicana de Investigacion y Docencia en Ingenierıa Quımica A.C. 543

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

Fig. 1: One-dimension sample reactor.

It implies that biomass is included on the carboncycle; hence the amount of CO2 produced from thedifferent transforming processes is balanced with theconsumption of this gas by the natural recyclingmethods. Among those processes, thermo-chemicalones represent interesting options due to its hightransformation yields, short reaction time, versatilityin relation with raw materials and byproducts, seeMorales-Martinez et al. (2014) and Evangelista-Flores et al. (2014). Pyrolysis is the process ofaccelerated degradation of a solid fuel (in this case,biomass) via an indirect heating source, in absenceof oxidizing agents; these operating conditions allowto condensate or synthesize specific compounds, suchas bio-oils and gases with high energy potential,for example methane and hydrogen. The use ofthese products as fuels is considered clean energybecause the byproducts of its combustion are balancedwith the carbon cycle. However, pyrolysis is acomplex process; it requires specific studies in order todetermine the best operating conditions to profit of itsadvantages. The application of the automatic controltheory could help to improve the operation protocolsand to enhance the process performances. Besides,a controlled environment on pyrolysis processesallows the minimization of undesired components orpollutant elements for post-processing and recyclingmethods. In this document, a fixed-bed wall-heatedpyrolysis reactor is studied. For this particular case,the degradation of the sample of wood is simplifiedas one-dimensional consumption process as shownin Figure 1, where the wood sample is exposedto the heat source on only one face, isolating therest of the sample to the external and neighboringreactions. Once the operational conditions wereidentified, a control strategy based on PI+SM controlwas designed for the regulation of the proposedheating path obtained in the dynamical analysis inorder to obtain a maximum rate of gas mixture.

2 Background

This wok focuses on the pyrolysis of wood wastes.Large amounts of this raw material are produced fromdifferent human activities; besides, the specializedliterature indicates that wood wastes present a regularstructure of its elemental components (50% carbon,42% oxygen, 6% hydrogen and 2% nitrogen andother elements). The thermal degradation mechanismof materials with vegetable origins has been wellidentified and documented by many authors. In nextlines, some related works are mentioned and brieflyanalyzed.

Grφnli (1995,1996) presents a variety ofmathematical models that satisfy the behavior of thepyrolysis process in an experimental environment, andthe necessity to analyze the heat dispersion at a particlelevel.

Basu (2006) explains that pyrolysis is a partialgasification because the process does not reach thetemperature of gasification (900 to 1700 K), and ahigh amount of liquid substance is created. This kindof gasification produces less volatile, generating highamounts of different kinds of solid residues.

Agblevor et al. (2010) sets the methodology andstrategies for the catalytic reaction of pyrolysis forobtaining selective products, using hybrid poplar woodin a fluidized bed reactor. In his experiments, Agblevorexplains that the selective products are the result ofa filtered condensation of steam and gases, with thismethod, the amount of char in the mixture decreases,making the oil residue more stable than the normal gasproduction of an ordinary pyrolysis process.

Mangut et al. (2006) characterizes the thermalproperties of the tomato by a thermogravimetricanalysis, in order to improve the efficiency ofcommercial pyrolysers, gasifiers and combustors. Thefuel used in the experimentation is tomato waste ina pulper-finisher machine. Once the obtained datawas analyzed by a mathematical model based onthe thermal decomposition of biomass pyrolysis, thekinetic parameters were optimized via a least-squaremethod.

Xue et al. (2011) proposes the use of Euler-Euler multiphase computational fluids dynamic forthe fast pyrolysis reaction of biomass in a fixed-bedreactor. The reactant fuel has variable porosity, theauthor uses this physical property to perceive thedynamical change of the degradation of the fuel as aresult of the different temperature exposures during thedevolatilization of the sample. The numerical solution

544 www.rmiq.org

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

proposed by Xue is based on the equality:

Biomass = Cellulose + Hemicellulose + Lignin (1)

The authors also separate kinetic parameters foreach species sub-products contained in the wood,generating a matrix composed of the individual kineticproperties of each bio-polymer, obtaining an accuratesimulation model.

Tsamba et al. (2006) says that the lignin startsthe degradation process at low temperatures (433 to443 K) and ends at 1173 K, meanwhile the cellulosestarts decomposition in the rank of 473 to 673 K.The degradation of these two cellular walls marksthe beginning of the decomposition of the solidsubstance. Tsamba also states that at over 673 K, thedecomposition rate of the solid substance diminishes.The experimental work consists in the use of cashewand coconut shells and the preparation of the samplesby a shredding process to obtain regular particle sizes.

Tingting et al. (2011) determines the experimentalcomposition of the gas and tar residues from thepyrolysis process. The gas mixture is composedmainly by CO, CO2, CH4 and H2, while theliquid mixture is composed of ketones, aldehydes,acids, esters, alcohols, alkenes, benzenes, phenols,carbohydrates and other biological residues. Theauthor also states that the carbohydrates and phenolsare a direct product of the decomposition of thecellulose, the main source of the phenols is the lignin,and the acids, ketones, aldehydes and phenols areproduced by the decomposition of xylose.

Di Blasi (2008) explains that the complicatedchain of reactions within pyrolysis process can besimplified by two main stages of the degradation: theprimary pyrolysis, at a range of 400 to 500 Celsius,which indicates the beginning of the degradation of thesolid substance in its two main components, gaseousphase and liquid residues (tarry mixture); and thesecondary pyrolysis, from 500 to 900 Celsius, whichstarts with the gasification of the liquid substance,mixing with the gaseous phase and increasing theamount of condensate volatile.

Westerhof et al. (2010), using pine wood ina fluidized-bed reactor with a feeding rate of 1 kgper hour, states that the production rate of the liquidspecies increases during the exposure to temperaturesbetween 330 and 450 Celsius, the rate remainsconstant between the 450 and 530 Celsius and this ratedecreases with temperatures above 580 Celsius.

As a thermodynamic process, a system withpyrolysis reaction is controllable. The importance ofcontrol in this type of process is mentioned by several

authors, Wu et al. (2012) reveals that the pyrolysisprocess has an elevated amount of uncertainty amongthe critical facts that enhance the production of thedifferent subspecies, and there is no reliable modelto predict these features. Wu proposed the utilizationof intelligent control (fuzzy control, neural networks,genetic algorithms and expert systems) for the controlof these kinds of reactions.

Likewise, Natarajan et al. (2009) proposed theuse of tubular fixed-bed reactor for the pyrolysis ofrice husk with wall heating from an external furnacecontrolled by a PID. This control maintains the reactortemperature between 400 and 600 Celsius. Theexperimental results of this author set that the particlesize is proportional to the production of the liquidresidue, this behavior responds to the improvement ofthe hydrocarbon cracking. The author explains that thelength increasing of the tube reactor raises the timeof residence, and this parameter decreases the finalproduction of liquid mixture.

The variable structure control is a relatively newmethodology for the pyrolysis process, Mingzhonget al. (2001) remarks that the methodology forthe implementation of a sliding mode control in anon-linear process by the setting of limits in thechattering phenomenon that is characteristic in thiskind of variable-structure control, improves the resultsof a controlled chemical reaction. This phenomenonconsists on a series of fast direction changes in thecontrol signal approaching the convergence to thesliding plane, and is induced by the fast response ofthe sliding mode control.

Hanczyc et al. (1995) described that thedistributed parameters in first-order non-linear partialdifferential equations can be simplified by a seriesof ordinary differential equations without the needof a parameter approximation, and apply geometricalmethods to use sliding mode control in this process.

According to several works few authors havedeveloped an experimental analysis to obtain productsof interest such as Agblevor et al. (2010) andWesterhof et al. (2010), and few others mentionthe control necessity to obtain favorable operationalconditions that enhance certain kinds of energeticproducts, such as gas, biodiesel or tar (see for instanceWu et al. (2012) and Natarajan et al. (2009)).

Therefore, in this work a feedback control for adistributed parameters pyrolysis reactor is proposed.The dynamic model is based on that one proposedby Grφnli (1995) and uses a heating source due thewall thermal transference, the energy balance equationwas uncoupled to its main parameters to improve

www.rmiq.org 545

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

computational costs; this is a common strategy to easethe study of complex processes, some other workshas been reported to deal with this situation, see forinstance Jimenez Islas et al. (2014). Once this solutionwas tested, a PI+SM control was implemented toregulate different heating paths.

The remainder of this work is organized asfollows: In section 3 the model and its numericalsolution are introduced, in Section 4 the PI+SMcontrol strategy and numerical results are presented,and finally, in Section 5 some conclusions arediscussed.

3 Model and numerical solutionBy the analysis of the kinetic and dynamic thermalproperties of the solid substance by Grφnli (1996),the degradation of the solid sample is represented byfour main stages as seen in Figure 2 where the solidsample kinetic transformation is represented by its3 main components, gas (k1), liquid (k2) and char(k3), which are the result of the primary pyrolysisdescribed by Di Blasi (2008), and as a secondaryproduct, the gasification of the liquid phase (k4). Thekinetic scheme leads to the sub-product transformationdynamics, defined by Grφnli and Melaaen (2000) as:

ωs = −ρs(k1 + k2 + k3) (2)

where ωs is the solid primary dynamicaltransformation rate and ρs is the initial density of thesolid substance. The gas mixture production rate (ωi)can be separated in its principal contributors, the gas

Fig. 2: Kinetic of the phase transformation.

(ωg) (3), and the liquid or tarry residues (ωT ) (4):

ω = ρsk1 + ρgT k4 (3)

ωT = ρsk2 − ρgT k4 (4)

By assuming thermal equilibrium between the solidand gas mixture phases (see Figure 1) and that thechanges in energy and mass are more important inthe x direction, the energy conservation equationdeveloped in Grφnli and Melaaen (2000) is given as:

ρggVgCpg

∂T∂x

+ (ρsCps + εgρggCpg)

∂T∂t

=

ke f f∂2T∂x2 −

∑ωi∆hi (5)

The first term is the convective heat transfer causedby volatile movement, and the second term representsthe accumulation of energy; the third term is theconductive heat transfer, which is represented by aneffective thermal conductivity; and the last term is thenet heat of reaction due the pyrolysis process.

In Figure 1, the heat source is considered in theboundary conditions as follows:

F f lux−hT (Ts−T∞)−ωSσ(T 4s −T 4

∞) = −ke f f∂T∂x

(6)

hT , ωs, σ are the convective heat transfer coefficient,surface emissivity, and Stefan-Boltzmann constant,respectively. Subscript ∞ denotes ambient and sdenotes surface. The pressure at the irradiated surfaceis assumed to be equal to the atmospheric pressure:

(Pg)gs = (Ts)∞g = Patm (7)

The most important assumptions of the mathematicalmodel (l)-(7) are: (a) all the phases are at the sametemperature; (b) the transport of bound water ismodeled as a diffusion process given by a diffusioncoefficient that is function of the local moisturecontent; (c) binary diffusion in the gas mixture; (d)Darcy’s law for the bulk flow of gas mixture andliquid water; (e) a linear variation between the virginwood and char for the properties related to the solidstructure; (f) wood shrinkage and crack formation arenot considered.

In equation (6) F f lux term represents the radiantheat source, which is considered constant for chosenperiods of time in Grφnli and Melaaen (2000). Inthe present work the heat source is considered to bevariable in order to match a desired temperature in thepyrolysis reactor. Then, the heating source affectingthe system is modeled by:

Q = Hwall(T ∗ −T ) (8)

546 www.rmiq.org

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

Fig. 3: Integration methodology for the open-looppyrolysis reactor.

with QF f lux as the variable heat applied through thewall, and is the result of the differences between thecontrolled temperature (T ∗), which represents a virtualcontrol input and is defined in the control section, andthe system temperature T , which for the simulationbecomes as Tk, i.e., the system temperature in thediscrete iteration k, see Figure 3; with a substantialloss through the wall heat transfer coefficient Hwall,see Figure 1, this wall in the schematic diagram ofthe experimental setup in Grφnli and Melaaen (2000),corresponds to the fused silica window that is directin contact with the variable-power xenon arc lampcapable of combustion-level heat fluxes that was usedas the source of energy to sustain pyrolysis. With theseassumptions the energy balance equation (7) can besimplified as can be seen in the following equations:

β = ρggVgCpg (9)

ξ = ρsCps + ερggCpg (10)

Therefore, the energy balance equation is defined asfollows:

β∂T∂x

+ ξ∂T∂t

= ke f f∂2T∂x2 −

∑ωi∆hi (11)

where the first part represents the convection heattransfer caused by the gas flow, the second part is theaccumulation of energy in the system via the specificheat of the sub-compounds, the third part is thethermal effective heat conductivity and the final partis the net heat production of the chemical degradation.To compute a numerical solution, the energy balanceequation can be uncoupled in its main parameters asshown in equations (12) and (13). The numericalvalues used in the solution are show in Tables 1

Fig. 4: Pyrolysis dynamic process in open-loop.

and 2, respectively:

β∂T∂x

=∑

ωi∆hi (12)

ξ∂T∂t

= ke f f∂2T∂x2 (13)

All the reaction dynamics for the three phases (liquid,gas and solid), and the heat produced by the reactionpaths related to time with the convective termsare solved by making a discrete integration with aconventional method to solve ordinary differentialequations. Meanwhile, for the conductive part, anapproximation of the solution is solved by a Taylorsecond order approximation (see Farlow, 1993):

∂2T∂x2 = f ′′(x) ≈

1h2 [ f (x + h)− 2 f (x) + f (x− h)] (14)

Discretization with the integration methodology isdepicted in Figure 3. Meanwhile, the behavior ofthe solid substance degradation without control isshown in Figure 4, which represents a realistic pathof degradation in comparison with the experimentalresults obtained by Grφnli et al. (2000), and fitsthe results reported by Di Blasi (2008) regarding thepyrolysis phases that this author found.

In a previous work Vazquez-Sandoval et al.(2012), the authors provided several trajectories thatwere tested using various types of conventionalcontrols. From this analysis, two main trajectorieswith the higher amount of gaseous substance obtainedby simulation were chosen. These trajectories werea ramp with two slopes and an exponential curve.To apply the regulation of these heating paths, thestarting point of the system was 300K, while the walltemperature of the heating source was 600K. Thewall temperature responds to a pre-heating process

www.rmiq.org 547

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

Fig. 5: Scheme of process plus PI+SM controller.

on the walls of the reactor on a temperature near theprimary pyrolysis starting point. The path equationsare defined by (15) for the ramp and (16) for theexponential path:

Tdk+1 = Tdk + x (15)Tdk+1 = Tdk + aex (16)

where α → 0 and x represents the temperatureincremental rate.

4 PI+ Sliding mode

With the sub-product analysis and the degradationresponse to the temperature presented in Vazquez-Sandoval (2012), the best options in the heating pathanalysis were selected for the control analysis. Forthe temperature control a PI+SM control strategywas used to regulate the system along the heatingtrajectory. Commonly for fast systems the PI+SMpresented in this document has been applied for thetemperature fast regulation, with a fast profile ofcontrolled temperature responding to certain desiredsequences of temperature rates. This kind of controlwas chosen because it guarantees the convergence ofthe system in finite time; this behavior minimizesthe difference between the desired temperature andthe system temperature from the beginning of thesimulation, improving gas production. This controloperates at the beginning of each iteration in thesystem as shown in Figure 5. This method increasesthe effectiveness of the solid transformation in gaseoussub-products against conventional control techniquesVazquez-Sandoval et al. (2012).

The PI+SM controller is defined by equations(17)-(19):

T ∗ = PI + S M (17)

PI = Kp(T −Td) + Ki

∫(T −Td) (18)

S M = λ(sat

(Sε

))+ T

where ε→ 0 (19)

T ∗ represents the controlled temperature of the heatingsource and is the feedback amount of temperatureadded to (8), T is the system temperature, and it isconsidered in the discrete iteration k, i.e., Tk; Td is thecorresponding temperature of the heating path in theinstant k, Kp and Ki are the proportional and integralgains respectively, meanwhile the sliding surface gainis defined by λ. The control gains are given by:

Kp = −1.7

Ki =Kp

7(20)

λ = −0.3

It is considered that the settling time for thetemperature is approximately 3 min for the higherconversion temperature, i.e. 900 C, accordingly toGrφnli and Melaaen (2000). With this in mind theproportional gain (Kp) is chosen approximately thehalf of this time (1.7), meanwhile, the integral gain(Ki) is selected near to the relation Kp ≥ 10Ki thatis recommended for process with first order dynamicresponse, which is the case of the temperature process(see Astrom and Hagglund, 1994). Finally, the gain λfor the SM is settled on top of the integral gain Ki toavoid instabilities.

To prevent the inherent chattering from the SMapproaching to the sliding surface, a saturationfunction of the sliding surface S is proposed.Simplified by ϕ = S/ε, ε = 0.1 was fixed on, thesolution of the system involves a conditional saturationfunction (see Khalil, 2002) with an hyperbolic tangentfor |ϕ| > 1 in order to obtain the approaching softer andprovide in S a continuous system defined by (21):

sat (ϕ) =

{ϕ |ϕ| 6 1

tanh(ϕ) |ϕ| > 1 (21)

S = Y + Y = 0 (22)

In (22) Y = T − Td, represents the regulation error andT is the approximation of the error derivative; and isobtained using a backward difference approximation

548 www.rmiq.org

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

(Farlow, 1993):

Y ≈(

Tk −Tk−1

∆k

)(23)

∆k is the discrete interval, Tk and Tk−1 are the currentand the former temperatures.

By deriving the sliding surface function (22) thefollowing equality is obtained:

Y = −Y (24)

For the nature of the sliding surface, as can be seen in(22), the error Y and its derivative T tend to 0, whilethe derivative of the sliding surface S satisfies theequality shown in (24), therefore this kind of systemcan be referred as asymptotically stable.

Figure 6 shows the reactor temperature result ofthe exposure to the heating ramp in closed loop withthe proposed control PI + S M (17) and under thePI control; meanwhile, the pyrolysis products curveunder the same ramp trajectory for PI + S M and PIcontrollers is depicted in Figure 7.

The fast response of the control exposes the solidsample to higher temperatures than the uncontrolledsystem; this phenomenon is produced because of theovershoot in the system response. This exposureleads to higher production rates for gaseous substance.Figure 8 shows the control action T ∗ along the time ofexposure to the ramp trajectory heating path.

Fig. 6: Control actions (PI+SM and PI) considering atemperature ramp trajectory.

Fig. 7: Products from pyrolysis with PI+SM and PIControllers under ramp a temperature ramp trajectory.

Fig. 8: Temperature control input T” added to theheating source along the time of exposure.

Comparing the PI + S M and PI under ramptrajectory, Figures 6, the error in the final stageof regulation diminishes with the SM action in thesystem, which means higher temperatures along thetime of exposure, and therefore, a higher amount ofgas conversion. The result for both control actions areshown in Table 3 also can be observed further in Figure9.

Figure 10 shows the pyrolysis products for thePI + S M controller when the system exposure to theexponential and the ramp paths. The exponentialtrajectory causes less profitable generation of gaseoussubstance, and the remaining primary solid substanceapproximates to zero. Figure 11 shows the gradientof temperature added by the control action (T ∗)along the time of regulation. Figures 12 and 13show the exponential heating path, under the controlaction of a PI controller without the SM action.

www.rmiq.org 549

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

Fig. 9: Zoom for control actions ramp trajectory.

Fig. 10: Pyrolysis products with PI + SM controllerunder exponential and ramp path.

Fig. 11: Control input T ∗ for the exponential path.

Fig. 12: Reactor temperature (T) with exponentialpath under PI control action.

Fig. 13: Pyrolysis products with PI control actionunder exposure to the exponential trajectory.

Fig. 14: Zoom in reactor temperature for PI + SM andPI controllers.

Fig. 15: Behavior of S against dS/dt.

In these figures, the system shows no overshootingat the first iterations and also shows a slowconvergence to the desired temperature, reducing thetime of exposure of the liquid phase to the heatingsource, resulting in a low gas count that can be seenin Table 3 and a closer view is shown in Figure 14.

In Figure 15, the graphic shows the plot of Sversus S , showing the behavior of the control signalapproaching to the convergence point at 0. EachX marks the signal S at the end of each iterationof the system. The first points at the left arethe chattering phenomena at the beginning of thesimulation approaching the sliding plane, ending in thepoint cloud that converges to 0.

The results displayed in Table 3 are a sample ofthe importance in the temperature control regulationof the predicted trajectory along the residence time ofthe solid sample. As Di Blasi (2000) explains, thesequence of the two stages of pyrolysis is necessary

550 www.rmiq.org

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

for the development of the process. In this work, thetrajectory described by the heating path allows thecontrol to guide the system through a pre-calculatedtime of exposure to the heating source, allowingthe final percentage of gas mixture to increase.This methodology is the basis for the proposal todesign a prototype reactor that will set the full-scalemethodology of selective production of high energeticsub-compounds.

Conclusions

In the pyrolysis process, temperature is a crucial factorin the final percentage of inert solid substance. Withthis premise, the control shown in this work based itsoperation on the variation of the heating source. Thisallows the modification of the maximum temperatureof the system. High temperatures (above 775 Kelvin)are necessary to promote a full transformation of thesolid substance.

According to literature, the temperature of thesystem is a primordial factor in the transformationof the solid substance in its main phases (Grφnliet al., 2000, Di Blasi, 2008). The control actionpresented in this document based its effectivenesson two main parameters: high temperature rising topromote the liquid phase creation during the primarypyrolysis, and high time of exposure at slow increasingtemperature to promote the gasification of the liquid atthe secondary pyrolysis stage, both in order to improvethe amount of final gas substance.

The control based on PI + S M shown to beeffective to obtain high gas production rates becausethe closed-loop pyrolysis process has a fast responseto the temperature changes in the prescribed trajectory.The next step is the experimental analysis of thecrucial conditions to promote the conversion of thesolid substance in selective sub-products such ashydrogen and methane, both highly energetic gaseswith direct applications as fuel for combustion motors.

As a future work, once the experimentalparameters have been settled, some optimization workneeds to be done in the field of trajectory planning ofthe heating process, such that this methodology canmaximize the results with an operational prototype ofa controlled pyrolysis reactor. Also, for this solutionhas to be considered and included both the efficiencyfactor for the heat added in the reactor (Q), and theefficiency of the chemical reactions. This would givea more accurate parameter of how the gas productionevolves and to take into consideration variation in the

residence time.

References

Agblevor, A., Beis, S., Mante, O., Abdoulmoumine,N. (2010). Fractional Catalytic Pyrolysis ofHybrid Poplar Wood. Industrial EngineeringChemical Research 49, 3533-3538.

Astrom, K., Hagglund (1994). PID Controllers:Theory, Design, and Tuning. Instrument Societyof America, 2nd Edition, North Carolina, USA.

Basu, P. (2006). Combustion and Gasification inFluidized Beds. Ed. Taylor and Francis, Florida,USA.

Di Blasi, C. (2008). Modeling Chemical andPhysical Processes of Wood and BiomassPyrolysis. Progress in Energy and CombustionScience 34, 47-90.

Evangelista-Flores, A., Alcantar-Gonzalez, F.S.,Ramırez de Arellano Aburto, N., Cohen Barki,A., Robledo-Perez, J.M., Cruz-Gomez, M.J.(2014). Diseno de un proceso continuo deproduccion de biodiesel. Revista Mexicana deIngenierıa Quımica 13, 483-491.

Farlow, S. (1993). Partial Differential Equations forScientist and Engineers. Dover Publications,New York, USA.

Grφnli, M. (1995). Experimental and ModelingWork on the Pyrolysis and Combustion ofBiomass- A Review of literature, SINTEFReport, STF12A95013.

Grφnli, M. (1996). Theoretical and ExperimentalStudy of the Thermal Degradation of Biomass.Ph.D. Thesis, Norwegian University of Scienceand Technology, Norway.

Grφnli, M., Melaaen, M. C. (2000). MathematicalModel of Wood Pyrolysis, Comparison ofExperimental Measures. Energy & Fuels 13,791-800.

Hanczyc, M., Palazoglu, A. (1995). Sliding ModeControl of Nonlinear Distributed ParameterChemical Processes. Industrial EngineeringChemical Research 34, 557-566.

www.rmiq.org 551

Morales-Diaz et al./ Revista Mexicana de Ingenierıa Quımica Vol. 14, No. 2 (2015) 543-552

Jimenez-Islas, H., Calderon-Ramırez, M., Molina-Herrera, F.I., Martınez-Gonzalez, G.M.,Navarrete-Bolanos, J.L., Castrejon-Gonzalez,E.O. (2014). Low-RAM algorithm forsolving 3-D natural convection problems usingorthogonal collocation. Revista Mexicana deIngenierıa Quımica 13, 259-277.

Khalil, H. (2002). Nonlinear Systems. Third EditionPrentice Hall, New Jersey, USA.

Mangut, V., Sabio, E., Ganan, J., Gonzalez, J. F,Ramiro, A., Gonzalez C. M., Roman, S., Al-Kassir, A. (2006). Thermogravimetric Studyof the Pyrolysis of Biomass Residues fromTomato Processing Industry. Fuel ProcessingTechnology 87, 109-115.

Morales-Diaz, A. Carlos-Hernandez, S. (2010).Saturated control design for bounded substrateregulation in a continuous anaerobic bioreactor.Revista Mexicana de Ingenierıa Quımica 9,219-229.

Morales-Martınez, T.K., Rios-Gonzalez, L.J., Aroca-Arcaya, G., Rodrıguez-de la Garza, T.K. (2014).Ethanol production by Zymomonas mobilisNRRL B-806 from enzymatic hydrolysates ofEucalyptus globulus. Revista Mexicana deIngenierıa Quımica 13, 779-785.

Mingzhong L., Fuli, W and Furong G. (2001). PID-Based Sliding Mode Controller for NonlinearProcesses. Industrial Engineering ChemicalResearch 40, 2660-2667.

Natarajan, E., Ganapathy Sundaram, E. (2009).Pyrolysis of Rice Husk in a Fixed Bed Reactor.World Academy of Science, Engineering andTechnology 56, 504-508.

Sacramento-Rivero, J.C., Romero, G., Cortes-Rodrıguez, E., Pech, E., Blanco-Rosete,S. (2010). Diagnostico del desarrollo debiorrefinerıas en Mexico. Revista Mexicana deIngenierıa Quımica 9, 261-283.

Tingting, Q., Wanjun, G, Laihong, S., Jun, X., Kun,Z. (2011). Experimental Study of Biomass

Pyrolysis Based on Three Major Components:Hemicellulose, Cellulose and Lignin. IndustrialEngineering Chemical Research 50, 10424-10433.

Tsamba, A., Yang, W, Blasiak, W. (2006). PyrolysisCharacteristics and Global Kinetics of Coconutand Cashew Nut Shells. Fuel ProcessingTechnology 87, 523-530.

Vazquez-Sandoval, A. (2012). Dynamical Analysisof a Pyrolysis Process for the Obtention ofGas with Energetic Potential. MSc Thesis,CINVESTAV, Saltillo, Coah. Mexico.

Vazquez-Sandoval, A., Morales-Diaz, A., Carlos-Hernandez, S. (2012). Temperature Controlof a Wood Pyrolysis Process. II InternationalCongress of the Mexican Academy ofInvestigation and Teaching in ChemicalEngineering, AMIDIQ, San Jose del CaboB.C.S. Mexico.

Vicencio-de la Rosa, M.G., Valencia-Vazquez,R., Perez-Lopez, M.E., Martınez-Prado, M.E.,Ramırez-Ramırez, E.S. (2015). Stabilizationby co-composting of solids removed fromwastewater treatment plant of a cheese factory.Revista Mexicana de Ingenierıa Quımica 14,61-71.

Westerhof, R.J.M., Brilman, D.W.F, van Swaaij,P.M., Kersten, RA. (2010). Effect ofTemperature in Fluidized Bed Fast Pyrolysis ofBiomass: Oil Quality Assessment in Test Units.Industrial Engineering Chemical Research 49,1160-1168.

Wu, J.H., Wang, J.K., Chen, J., Si, H.P, Lin, K.Y(2012). The Control of Fixed-Bed Reactors forBiomass. Advanced Materials Research 424-425, 1117-1123.

Xue, Q., Heindel, T. J., Fox, R. O. (2011). A CFDModel for Biomass Fast Pyrolysis in Fluidized-bed Reactors. Chemical Engineering Science66, 2440-2452.

552 www.rmiq.org