Baixe o app para aproveitar ainda mais
Prévia do material em texto
Chapter 4 Reactive Systems When reactors are present in the flowsheet under consideration, there are more choices that must be made regarding how variables are counted and how the balances are formulated. In previous chapters a pattern was established of beginning with a section on tools and techniques and then presenting collections of sample flowsheets in which the application of the new tools is illustrated. That pattern is continued here. Since the number of new tools and techniques required to deal with reactive systems is extensive, section 4.1 consists of five sub-sections. Section 4.1.1 introduces the additional terms and tools necessary to analyze flowsheets with reactors. Terms defined here include stoichiometric coefficient, reaction rate, fractional conversion, limiting reactant, stoichiometric air, and excess air fraction. The formalism for dealing with more than one reaction is developed in section 4.1.2. When multiple reactions occur a fractional yield may be given as a specification. The issue of reaction independence also arises. Element balances find use under special circumstances explained in section 4.1.3. The concept of the atom matrix is introduced along with the question of the independence of the element balances. In section 4.1.4 a technique is introduced which permits the identification of an independent set of reactions given only the atom matrix for a system. This technique is referred to as the stoichiometric procedure. Lastly, the issue of dependence in reactor specifications is presented in section 4.1.5 along with a method for detecting combinations of reactor specifications which do not represent independent information. Section 4.2 is a collection of flowsheets in which a single reaction occurs, with and without recycle. The flowsheets in section 4.3 contain a single reactor but multiple reactions occur. The focus is on the formalism needed to deal with material balances for such systems. Section 4.4 is a collection of problems in which element balances are effectively used. Some issues dealing with fossil fuels are also presented. In section 4.5 several flowsheets are analyzed in which no reactions are given for reactors with a complex set of products in the exit stream. The stoichiometric procedure is applied in each of these systems to obtain an independent set of reactions. 4.1 Fundamental Tools 4.2 Single Reactions 4.3 Multiple Reactions 4.4 Unspecified Stoichiometries 4.5 Formal Stoichiometries 4.1 Fundamental Tools A generalized approach to stoichiometry makes available a set of tools that are useful in dealing with reacting systems. This section summarizes the techniques needed to count and to formulate meaningful species or element balances, and to interpret specifications that are unique to reacting systems. 4.1.1 Single Reactions 4.1.2 Multiple Reactions 4.1.3 Element Balances 4.1.4 Generating Independent Reactions 4.1.5 Dependence in Reactor Specifications 4.1.1 Single Reactions The terms stoichiometric coefficient, reaction rate, conversion, excess fraction and limiting reactant are defined here. Topics related to the combustion of fuels are discussed, in particular the concepts of an ultimate analysis, stoichiometric air ratio and excess air fraction. The counting of variables is reviewed for the special case of a single reaction. Stoichiometric Coefficient The coefficients in a balanced chemical reaction are referred to as stoichiometric coefficients. Any chemical reaction among S components can be symbolized as 0 1 S s σs Cs⋅∑ = = where Cs is the chemical symbol for species s. For example, the reaction to form ammonia is N2 + 3H2 = 2NH3. It can be also be written 0 2 NH3⋅ N2− 3 H2⋅−= You will note that the coefficients for products are positive, while those for reactants are negative. For the above reaction: σNH3 = 2, σN2 = -1, σH2 = -3. Reaction Rate Let the molar flowrates of species s into and out of a reactor be ns,0 and ns,1 respectively. Then the reaction rate r is independent of the species selected and is defined as r ns 1, ns 0,− σs = For example: We can compute the rate of the reaction using any species for which information is known in both reactant and product streams. Using hydrogen: r = [(28 - 40) mol/h] /(-3), nitrogen: r = [(8 - 12) mol/h]/(-1) or ammonia: r = [(8 - 0) mol/h] /(+2) the same reaction rate r = 4 mol/h is obtained. Limiting Reactant The reactant species for which the quantity [ ns,0 / (-σs) ] is a minimum is called the limiting reactant. For the reactor above, there are two reactants. [ nH2,0 /(-σH2)] = 40/3 while [ nN2,0 / (-σN2) ] = 12/1, so that for the feed shown, N2 is the limiting reactant. Note that the term "limiting reactant" is only defined for a fully specified set of feed streams. Fractional Conversion The term fractional conversion applies only to reactants. Using inlet and exit streams numbered 0 and 1 respectively, the fractional conversion of species s is defined as Xs ns 0, ns 1,− ns 0, = For example, the fractional conversion of H2 is XH2 40 28− 40 0.3=:= The fractional conversion of N2 is XN2 12 8− 12 1 3 =:= Multiple Feed, Product Streams If a reactor has more than one feed stream and several product streams the definitions presented above can be unambiguously extended. For example, let the ammonia formation reaction occur N2 + 3 H2 = 2 NH3 in a reactor with I feed streams and J product streams numbered as in the figure above. Then the definitions of reaction rate, fractional conversion, and limiting reactant become: r I 1+ I J+ i ns i,∑ = 1 I i ns i,∑ = − σs = s = H2, N2, NH3 Xs 1 I i ns i,∑ = I 1+ I J+ i ns i,∑ = − 1 I i ns i,∑ = = s = H2, N2 The limiting reactant is the reactant species for which the following quantity is a minimum. 1 I i ns i,∑ = σs− s = H2, N2 Combustion of Fuels Many industrial processes involve the combustion of a fuel, usually with air. Many different methods for designating the amount of air used have been devised. These are reviewed here, first for simple fuels consisting of a single, well-defined species and then for complex fuels such as the fossil fuels: coal, oil. CH4 Combustion The stoichiometric amount of oxygen required to burn a fuel is defined to be that amount necessary to convert all carbon in the fuel to CO2, all hydrogen to H2O and all sulfur to SO2. If the fuel contains any nitrogen, it is assumed to leave as N2 in the product stream. Note, that this definition says nothing about the reactions that actually occur in the combustion process. The "combustion" reaction is that defined above. For example, the combustion reaction for methane is CH4 + σ O2 = CO2 + 2 H2O Notice that the coefficients for CO2 and H2O are determined by the molecular formula for methane. The stoichiometric coefficient for oxygen is determined by an O-balance. In this case σ = 2. Stoichiometric Oxygen Now consider a fuel with an elemental composition given by a set of atomic coefficients α. Let the elements occur in the order C, H, S, O, N. So that the fuel has the formula Cα0 Hα1 Sα2 Oα3 Nα4. The balanced combustion reaction for one "mol" of fuel is then Fuel + σ O2 = α0 CO2 + ( α1/2 ) H2O + α2 SO2 + (α4/2) N2 The coefficient for oxygen is obtained by an O-balance. σ 2 α0 α2+( )⋅ α12+ α3− 2 = Stoichiometric Air Ratio The molar amount of air required per mol of fuel is simply the stoichiometric coefficient of oxygen in the combustion reaction divided by the concentration of oxygen inair. The stoichiometric air/fuel ratio β is usually placed on a mass basis β fair stoich ffuel = nair( )stoich nfuel Mair Mfuel ⋅= σ xO2 air, Mair Mfuel ⋅= Ultimate Analysis Many substances are used in combustion equipment including fossil fuels such as coal, oil, and "natural" gas as well as many agricultural residues such as waste wood, corn stover, bagasse, peanut shells etc. Municipal solid waste is often burned to recover energy. These substances are typically characterized by their elemental analyses on a mass basis. Chemical Formula To convert the elemental analysis to a chemical formula, simply divide the mass fraction of each element by its atomic weight. This yields the relative molar amounts of each element. Only the ratios of these quantities are significant. To obtain a chemical formula, it is customary to select one of two bases: (1) a fixed molecular weight or (2) a specified coefficient for carbon, often six. This simply means scaling all the coefficients to give carbon a coefficient of exactly six. Both of these bases are illustrated in the following examples. Mair 28.97 gm mol ⋅:=Molecular weight of air xO2 0.21:=Oxygen concentration in air αC 6:=Desired coefficient of carbon A 12.011 1.008 32.064 16.00 14.007 gm mol ⋅:=η 0.8400 0.1140 0.0320 0.0000 0.0140 := Atomic weightsUltimate analysisData: αeCoefficient of element e in chemical formula for oil e 0 4..:=Element indices (C, H, S, O, N)Define: A fuel oil has the chemical composition: C: 84.00 wt%, H:11.40%, S: 3.20%, N: 1.40%. Determine a formula for the oil on a C-6 basis and the stoichiometric amount of air to burn this oil. Example 4.1.1.1 Moil 100 gm mol =Moil e αe Ae⋅∑:= α 6.994 11.31 0.1 0 0.1 =αe αe M'oil Moil ⋅:= The elemental coefficients calculated need only be scaled to match the desired formula weight. M'oil 100 gm mol ⋅:=Data: Repeat the above calculations but use a basis of a specified molecular weight for the oil. Example 4.1.1.2 Thus 13.686 g-air is required to burn one g of fuel. β 13.686=β σ xO2 Mair Moil ⋅:=Stoichiometric air/fuel ratio: σ 8.511=σ 2 α0 α2+( )⋅ α12+ α3− 2 :=Stoichiometric oxygen: Moil 85.793 gm mol =Moil e αe Ae⋅∑:= Since the coefficient of C was set, the molecular weight of the oil is also determined by this choice. That is, the chemical formula for the oil is C6 H9.703 S.086 N.086. α 6 9.703 0.086 0 0.086 =αe ηe η0 A0 Ae ⋅ αC⋅:= Chemical formula for oil:Example 4.1.1.1 (continued) nair nair stoich 1 Φ+( )⋅= σ xO2 1 Φ+( )⋅= Given an excess air fraction Φ, the actual amount of air used is then computed using the stoichiometric amount of air for that fuel: Note that an excess air fraction of 0.0 means that a stoichiometric amount of air was used, while an excess air fraction of 0.1 means "10% excess air", and a fraction 1.2 means "120% excess air". Φ nair nair stoich 1−= A common specification for combustion processes is the "percent excess air". We will use the term excess air fraction defined as When burning fossil fuels or agricultural waste materials to generate energy, it is usual practice to use more than the stoichiometric amount of air. The excess oxidant promotes a more intense combustion reaction and makes it possible to achieve complete conversion of the fuel. Excess Air Fractions As before, 13.686 g-air is required to burn one g of fuel. β 13.686=β σ xO2 Mair Moil ⋅:=Stoichiometric air/fuel ratio: σ 9.921=σ 2 α0 α2+( )⋅ α12+ α3− 2 := Stoichiometric oxygen: The stoichiometric coefficient of oxygen in the combustion reaction has changed because the formula for the oil has changed, but the air fuel ratio is the same. n 40 12 0 30 8.667 6.667 mol hr = ns 1, ns 0, σs r⋅+:=Species balances: n1 0, 12 mol hr ⋅:= σ 3− 1− 2 := n2 0, 0 mol hr ⋅:=n0 0, 40 mol hr ⋅:= s 0 2..:=Let the species be indexed in the order H2, N2, NH3: Now all the product flowrates can be computed from the species balances. r 10 3 mol hr ⋅:=r XH2 n0 H2,⋅ σH2− := The DF analysis indicates that there is sufficient information to solve this problem. The definitions of reaction rate and conversion are combined to give Item Reactor __________________________ Variables Stream 5 Rxn rate 1 -------------------------- Sp. Balances 3 Flowrates 2 Conversion 1 __________________________ DF 0 In the reactor shown below, the conversion of H2 is 0.25. Compute all the flow rates. Each independent reaction adds one reaction rate variable to the total number of variables for the flowsheet. If a single reaction is occurring then of course, only one reaction rate is needed. Counting Variables mol mole≡ 4.1.2 Multiple Reactions When multiple reactions occur there is a need for a formal approach to keep track of the stoichiometry. The stoichiometric matrix is defined for this purpose. When counting variables, a number of reaction rates equal to the number of independent reactions must be included. The number of independent reactions is simply the rank of the stoichiometric matrix. These issues are explored in this section. When multiple reactions occur, an additional type of specification arises: fractional yield. The definition of this quantity Stoichiometric Matrix The stoichiometric coefficient of species s in reaction k is σs,k. This two-index quantity is the stoichiometric matrix for the reacting system. The extension of the species balances to multiple reactions is straightforward. Each species is augmented by its weighted participation in each reaction that occurs. For input stream 0, exit stream 1 of a reactor in which R reactions occur with rates rk, we have ns 1, ns 0, 1 R k σs k, rk⋅∑ = += The use of the stoichiometric matrix in performing material balances is illustrated in the next example. Example 4.1.2.1 One hundred mols/hr of an equimolar mixture of A and B reacts with 20% conversion of A to produce a mixture of A,B,C,E in which the concentration of C is 1.5 times that of E. There are only two reactions occurring: A + B = 2C and B + C = E. Stream Vars: 6 Equipment Vars: 2 --------------------- Balances: 4 Flows: 2 Conversion 1 Comp. ratio 1 _____________________ DF 0 It is clear that we have sufficient information to proceed. Try doing the material balances on your own. Click on the text below to check your work. View the material balances Independent Reactions In many systems, particularly at high temperatures, the precise pathway by which reactants are converted to products is unknown or extremely complex. These are vital issues for reactor design. But for material balance purposes it is usually satisfactory to represent the overall chemistry by some list of simple reactions. This list must be complete enough to account for all products. For the purposes of counting variables, the list of reactions must also be independent. That is, no reaction can be a sum or difference of any other subset of reactions. The precise number of such reactions needed to represent the stoichiometry of any system can be determined from the stoichiometric matrix. Let the number of nonzero columns obtained after exhaustive column reduction of σ be symbolized by ρ(σ). Wecall this the rank of σ. In performing a DF analysis for a multi-reactive system , we count exactly ρ(σ) reaction rate variables. Example 4.1.2.2 The steam reforming of methane results in the following reactions. CH4 + CO2 = 2CO + 2H2 CO + H20 = CO2 + H2 CH4 + H20 = CO + 3H2 CH4 + 2H2O = CO2 + 4H2 There are five species (CH4,CO2,CO,H2O,H2) and four reactions, so the stoichiometric matrix is 5 x 4. σ 1− 1− 2 0 2 0 1 1− 1− 1 1− 0 1 1− 3 1− 1 0 2− 4 := σ 1 0 1− 1 3− 0 1 1− 1− 1 0 0 0 0 0 0 0 0 0 0 := The second form is obtained from the first by elementary column operations. These include multiplying by a constant, or adding a multiple of one column to another. The final form shows clearly that there are only two independent reactions for this system, that is, r(s) = 2. The reactions represented by the second form of the stoichiometric matrix are: CO + 3H2 = CH4 + H2O and C0 + H2O = CO2 + H2. For many systems it is easy to determine by inspection that a set of reactions is independent. If each reaction contains one unique species then each reaction is necessarily independent. In more complex cases, the matrix reduction procedure may be required. When a feed substance is converted to multiple products by more than one reaction, it is common to specify the relative amount of some desired product obtained. This new type of specification is called the yield. Yield This ratio applies only to reactors with at least two competing reactions. Let reactant A form desired product C by one reaction and one or more other products (but no C) by competing reactions. The yield of C is simply the actual amount of C formed divided by the maximum amount of C that would have been formed if all the A converted went to producing C. Again using streams 0, 1 as inlet and exit streams: YC σA nC1 nC0−( )⋅ σC nA1 nA0−( )⋅= The s's are the stoichiometric coefficients in the balanced reaction that produces C. Recall that such coefficients are defined to be negative for reactants, so that the ratio above is always positive. 4.1.3 Element Balances For most systems, the species balances give rise to algebraically simpler equation sets than do the element balances. The material balances for a special class of problems, however, can be simpler when phrased in terms of element balances. This section explores the issues involved in setting up and solving element balances. Elements are Conserved In reacting systems, species are not conserved and the balance equations must reflect this. Recall that the species balances for reacting systems have terms involving the stoichiometric matrix and reaction rates. In contrast, elements are conserved. But the balances do not reduce to the unrestricted form of the non-reactive species balances. The elements are "packaged" into distinct species, and our material balance variables are the flowrates of these species. The balances must be expressed in terms of species flowrates but reflect element conservation. Atom Matrix The formal method of dealing with this is to define an atom matrix. The element αes of this matrix is simply the coefficient of element e in the molecular formula for species s. The element balances merely tally the amount of each element accompanying each species entering and leaving the system. If there are I input streams, and J exit streams in a system containing S species consisting of E elements, then the balances are 1 J j 1 S s αe s, ns j,⋅∑ = ∑ = 1 I i 1 S s αe s, ns i,⋅∑ = ∑ = = e = 1 ... E There is one such equation for each element. The formidable double-sum form of these balances hides an underlying simplicity: each element is conserved. In practice, the equations are very simple to formulate, as shown in the next example. First, notice how many terms are involved in the element balances even for a unit with only one inlet and one exit stream. The balances are very easy to write down, because element conservation is a simple concept. But the equations are usually algebraically less tractable than the corresponding species balances because typically many terms appear in each equation. Things to Note About Ex4.1.3.1 nH2O 41 15 mol hr ⋅:=nCH4 2 15 mol hr ⋅:=nCO2 21 15 mol hr ⋅:=nCO 7 15 mol hr ⋅:= Substituting nCO2 = 3 nCO and solving the resulting system of three unknowns by elementary methods we obtain the solution: 2 nCO2⋅ nCO+ nH2O+ 2 nO2⋅=Oxygen nH2O 2 nCH4⋅+ 3 nC2H6⋅=Hydrogen nCO2 nCO+ nCH4+ 2 nC2H6⋅=Carbon DF = 6 - (3 + 2 + 1) = 0 where we have three element balances, two flowrates, and one compostion ratio. The problem is well posed. Note that we do not count any reaction rate variables when doing element balances. The element balances are Ethane is burned with oxygen to produce carbon dioxide, carbon monoxide, methane, and water. When a 3:1 ratio of oxygen to ethane is used, the ratio CO2/CO in the product is 3:1. Example 4.1.3.1 Independent Element Balances If there are E elements in a system, there will be E element balances that can be written. Ordinarily, these will be independent. But for some systems this isn't true. This occurs frequently enough that a method of counting independent balances is needed. The number of independent element balances for a system with atom matrix α is simply ρ(α), the rank of the atom matrix. The rank of the atom matrix is the number of nonzero rows remaining after exhaustive row reduction. The technique is illustrated in the next example. Example 4.1.3.2 Urea is produced by reacting ammonia with carbon dioxide. When a feed consisting of 1/3 CO2 is reacted with 99% conversion of the carbon dioxide, the product stream contains the species shown in the figure. Is this problem well-posed? The problem can be well-posed only if there are four independent element balances. DF = 7 - [ ρ(α) + 1 + 1 + 1] Numbering the species in the order listed for the product stream, and numbering elements in the order (C,O,H,N) the atom matrix is α 1 1 4 2 0 1 2 0 1 2 6 2 1 2 0 0 0 0 3 1 := Elementary row operations can be used to reduce the atom matrix to a form in which there is a zero row. Then ρ(α) = 3 and the problem is underspecified from this viewpoint. α 1 0 0 0 0 1 0 0 1 1 0 0 1 1 2 0 0 0 1− 0 := Summary In general, a good rule of thumb is to always use species balances for non-reactive systems. The number of independent species balances is always S, the number of species, and S >= ρ(α) for any system. Also, the species balances are usually easier to solve. For reactive systems, when ρ(σ) = S - ρ(α), it is feasible to use either species or element balances. When this condition does not hold, use species balances. In other words, use element balances only when they offer the possibility of an obvious simplification. Element balances are only essential when a reactant is a complex mixture such as a fossil fuel or agricultural product such as wood, corn stover, or bagasse. 4.1.4 Generating Independent Reactions Species balances are preferable to element balances in most cases. But what if there are no reactions given? Suppose that all we know about the system is the set of species present. The species balances require that we have a set of independent reactionsin order to formulate meaningful expressions for the product flow rates. This section discusses how to generate such a set of reactions knowing only the atom matrix. The resulting technique is called the stoichiometric procedure. Stoichiometric Procedure: α ==> σ Stoichiometric Procedure With knowledge only of the species present in the system we can write the atom matrix α for the system. This is an array with E rows and S columns, where E is the number of elements present and S is the number of species. To determine a set of independent reactions for this system, use elementary row operations to put α in the form α I 0 β 0 = I is an identity matrix of size ρ(α), and β has dimensions ρ(α) x [S - ρ(α)]. The stoichiometric matrix is then an S x [S - ρ(α)] matrix of the form σ β− I = The next example illustrates the procedure. Example 4.1.4.1 This "alphabet soup" system consists of five species A,B,C,D,E and four elements w,x,y,z. The molecular formulas of the species are reflected in the atom matrix shown below. A = w x y3 B = y2 z C = w y D = w2 y4 z E = x y2 The system consists of the species: It is represented by the atom matrix: α 1 1 3 0 0 0 2 1 1 0 1 0 2 0 4 1 0 1 2 0 = The atom matrix can be row-reduced to the form: α 1 0 0 0 0 1 0 0 0 0 1 0 0 1 2 0 1 0 1− 0 = We identify a 3x3 identity matrix in the upper left, and the matrix β is: β 0 1 2 1 0 1− = The stoichiometric matrix is obtained by taking the negative of b and appending an identity matrix of size [S - ρ(α)] or 2. σ 0 1− 2− 1 0 1− 0 1 0 1 = This matrix corresponds to the two reactions: B + 2C = D and A = C + E. Additional examples are given by the ammonia oxidation system of section 4.5.1 and the alcohols from syngas system of section 4.5.2. Summary When a system is characterized by knowledge only of the species present, a set of independent reactions can easily be found by the stoichiometric procedure. 4.1.5 Dependence in Reactor Specifications Once a set of R independent reactions has been specified for a reactor, information given in the specifications is then used to fix the reaction rates. Fixing a reaction rate requires that at least one species be completely specified: that is, both its input and output flow rates must be known. Some combinations of completely specified species do not represent independent information. In this section a test is presented to detect when this situation occurs. Example Let the combustion of methane with pure oxygen be modeled with the following two reactions: CH4 + 2 O2 = CO2 + 2 H2O 2CH4 + 3 O2 = 2CO + 4 H2O The methane is completely burned to produce an oxygen-free stack gas consisting of CO2, CO, and H2O. The two reactions are clearly independent since each involves a unique species. There are five balances and a basis can be selected. We will use the methane flow rate. This means that one additional specification can be made. The same conclusion is drawn from a DF analysis. DF = [5 + 2] - [5 + 1] = 1 Let this additional specification be the water flow rate. The two species that are completely specified are methane and water. Let the methane flow rate be 1.0 mol/h and that of water be 2.0 mol/h. Their balances are CH4: 0 = 1 - r1 - 2 r2 H2O: 2 = 2 r1 + 4 r2 By an obvious rearrangement, the water balance is seen to be a multiple of the methane balance, and is therefore not an independent specification. Criterion for Independence For species balances, a specification is independent if the rows of the stoichiometric matrix corresponding to the completely specified species are linearly independent. In the above example, the stoichiometric matrix with species ordering methane, oxygen, water, carbon dioxide, carbon monoxide is: σ 1− 2− 2 1 0 2− 3− 4 0 2 = The rows corresponding to the completely specified species ( methane and water ) are clearly not linearly independent. The row for water is just (-2) times the row for methane. σcs 1− 2 2− 4 := Note that the original problem can be solved if another specification is selected. For example, a flowrate of CO2 or CO, or a CO2/CO ratio would be valid specifications. 4.2 Single Reactions Each flowsheet in this section involves a single reaction. The sample problems are intended to give you practice in setting up a stoichiometric matrix and using an assortment of specifications to find the product composition. 4.2.1 Chlorine Dioxide Synthesis 4.2.2 Hydration of Ethylene 4.2.3 Acetic Acid Synthesis 4.2.1 Chlorine Dioxide Synthesis Problem Statement Chlorine dioxide is a bleaching agent produced by the following reaction. 6 NaClO3 + 6 H2SO4 + CH3OH = 6 NaHSO4 + 6 ClO2 + CO2 + 5 H2O If 14 mols of the chlorate stream are fed for every mol of methanol and a conversion of 90% is achieved, what is the composition of the product? Click on the icon at the left to see the DF analysis for this problem. Click on the underlined text below to see the material balances. View the material balances for the chlorine dioxide synthesis. 4.2.2 Hydration of Ethylene Problem Statement The catalytic hydration of ethylene is carried out with a single-pass conversion of only 4.5%. The separator removes all the alcohol and the recycle gas contains 6.5 mol% steam. The molar ratio of steam to ethylene in the mixed feed to the reactor is 0.55. Calculate the compositions of all streams, and the ratio of steam to ethylene fed to the process. On Your Own Click on the underlined text below to check your DF analysis. View the DF analysis for the hydration of ethylene. Complete the material balances for this problem, then check your work by clicking on the underlined text below. View the material balances for the hydration of ethylene. 4.2.3 Acetic Acid Synthesis Problem Statement The recycle reactor shown below oxidizes ethanol to acetic acid with a mixture of dichromate and sulfuric acid. A 90% overall conversion is achieved with a recycle to ethanol feed ratio of 1:1, a 10% excess of sulfuric acid, and a 20% excess of sodium dichromate. The recycle stream contains only ethanol and sulfuric acid. The concentration of acid in the recycle is 94 mol%. A perfect separation of acetic is achieved. The reaction occurring is 3C2H5OH + 2NaCr2O7+ 8H2SO4 = 3CH3COOH + 2Cr2(SO4)3 + 2Na2SO4 + 11H2O On Your Own Perform a DF analysis on the acetic acid synthesis flowsheet. Check your work by clicking on the underlined text below. View the DF analysis for the acetic acid synthesis Now do the material balances in your worksheet. Check your work by clicking on the underlined text below. View the material balances for the acetic acid synthesis. 4.3 Multiple Reactions The flowsheets in this section all involve multiple reactions. These problems focus on calculating the reaction rates from given specifications and then using those rates to compute final product compositions. 4.3.1 Acetaldehyde Synthesis 4.3.2 Carbon Disulfide Synthesis 4.3.3 Reformate Hydrodealkylation 4.3.4 Toluene Dealkylation 4.3.1 Acetaldehyde Synthesis Problem Statement Acetaldehydecan be produced by the catalytic dehydrogenation of ethanol. There is however, a parallel reaction that produces ethyl acetate. C2H5OH = CH3CHO + H2 2C2H5OH = CH3COOC2H5 + 2H2 In a particular reactor, an ethanol conversion of 95% is obtained with an 80% yield of acetaldehyde. Calculate the product composition assuming the feed consists of pure ethanol. On Your Own Try performing the DF analysis for this problem. Check your work by clicking on the exercise icon at the left. Now compute the product composition. Click on the underlined text to check your work. View the Acetaldehyde Synthesis material balances. 4.3.2 Carbon Disulfide Synthesis Problem Statement Vaporized sulfur is reacted with methane to produce CS2. There are three reactions known to occur. CH4 + 4S = CS2 + 2H2S CH4 + 2S = CS2 + 2H2 CH4 + 2H2S = CS2 + 4H2 A feed containing four mols of sulfur per mol of methane results in a product mix characterized by 90% methane conversion and 70% sulfur conversion. Calculate the product composition. On Your Own Complete a DF analysis for this problem. Click on the underlined text below to check your work. View the DF analysis for the Carbon Disulfide synthesis. Do the material balances for this problem. Click on the underlined text below to check your work. View the material balances for the Carbon Disulfide synthesis. Click here to check your DF analysis for the hydrodealkylation Determine if the problem is well-posed.On Your Own What is the composition of the product stream? (4)2C6H6 = (C6H5)2 + H2 (3)C6H3(CH3)3 + H2 = C6H4(CH3)2 + CH4 (2)C6H4(CH3)2 + H2 = C6H5CH3 + CH4 (1)C6H5CH3 + H2 = C6H6 + CH4 The reactions occurring are: The process of removing alkyl groups from aromatic compounds with hydrogen is called hydrodealkylation. In the process shown below, a refinery reformate stream consisting of 5 mol% benzene (B), 20% toluene (T), 35% xylene (X), and 40% mesitylene (Y) reacts with hydrogen. When 5.0 mols of hydrogen is used per mole of hydrocarbon feed, the following conversions are obtained: T:80%, X:74%, Y:13%. The product stream also contains 0.1% biphenyl (P). Problem Statement 4.3.3 Reformate Hydrodealkylation Compute the composition of the product stream. Click here to see the material balances for the hydrodealkylation 4.3.4 Toluene Dealkylation Problem Statement The dealkylation of toluene to produce benzene is accompanied by a side reaction which produces biphenyl. C6H5CH3 + H2 = C6H6 + CH4 2C6H5CH3 + H2 = (C6H5)2 + 2CH4 The side reaction requires that conversions be kept low in the reactor, and this complicates the separation sequence. The first stage of separation recovers pure benzene and produces two recycle streams. The overhead consists of hydrogen and methane, part of which must be purged to remove the methane. The bottom product of the first separation stage is a liquid stream which undergoes further separation into a pure toluene recycle stream, and a pure diphenyl byproduct. When the reactor feed has a 5:1 ratio of hydrogen to toluene, a 75% toluene conversion is achieved. a. If the reactor outlet stream is 5% benzene, and 2% toluene, calculate the fractional yield of benzene and the make-up (stream 2) hydrogen required per mol of toluene fed to the process. b. Repeat for a reactor outlet specification of 2% toluene and 58% methane. On Your Own In your worksheet, perform a degree of freedom analysis for part a. Click on this text to check your DF analysis. Continuing in your worksheet, compute all the material flows. Click on this text to check your material balances. Now repeat the DF analysis using the specifications for part b. Click on this text to check your DF analysis for part b. 4.4 Unspecified Stoichiometries The problems in this section focus on the application of element balances to process flowsheets. Species balances are usually preferred over element balances. Element balances can be useful, however, in processes involving natural products or in cases where the reaction stoichiometry is unspecified. Natural products such as fossil fuels or agricultural residues are complex mixtures of materials with a wide range of molecular weights. Their composition as a mixture of distinct species is too complex to deal with effectively for material balance purposes. Instead, balances are done at the element rather than the species level. The reactions involved in some processes, even those involving only simple, distinct species, are often unknown or very complex. Element balances are sometimes useful for these systems. 4.4.1 Fuel Oil Combustion 4.4.2 Coal Gasification 4.4.3 Fuel Oil Desulfurization 4.4.4 Formaldehyde Synthesis 4.4.1 Fuel Oil Combustion Problem Statement A fuel oil with analysis C:84.0%, H:11.4%, N:1.4%, S:3.2% is burned with air. The stack is found to contain 2130 ppm by wt (dry basis) of sulfur dioxide. Determine a chemical formula for the oil on a C-6 basis, and compute the percent excess air that was used in the burner. On Your Own Perform a DF analysis for the oil burner. Click on the this text to see the DF analysis for the oil burner. On your worksheet, solve for the percent excess air. Click on this text to view the material balances for the oil burner. 4.4.2 Coal Gasification Problem Statement A dry coal consisting of ash:3%, N:1.5%, S:0.6% and unknown amounts of C, H, and O is gasified with steam and air to produce a fuel gas. The dry-basis composition of the fuel gas is H2:23%, CH4:3.2%, CO:16.2%, CO2:12%, H2S:1%, and N2:44.6%. In addition to the fuel gas, a solid residue is produced consisting of 10% C and the rest ash. If a 1:1mass ratio of steam/coal is fed, calculate the air rate and moisture content of the fuel gas. On Your Own Perform the DF analysis for this system. Click on this text to check your DF analysis for the coal gasifier. Now complete the material balances. Click on this text to see the material balances for the coal gasifier. 4.4.3 Fuel Oil Desulfurization Problem Statement A fuel oil with analysis C:84.0%, H:11.4%, N:1.4%, S:3.2% is desulfurized by contact with a hydrogen-rich gas ( 44 mol% H2, 56% CH4) in a catalytic reactor. The desulfurized oil contains only 1 ppm N, 9 ppm S. The gas stream exiting the reactor contains H2, CH4, NH3, H2S. The methane to hydrogen ratio in the off-gas is 7:3, while the ammonia:hydrogen ratio is 1:4. The reactor off-gas is subjected to two stages of purification. The first stage removes 90% of the hydrogen sulfide and all of the ammonia. The second stage removes all the remaining H2S. The recycle stream is 40% H2. It is mixed with fresh gas containing 48% H2. All gas compositions are in mol %, while the oil compositions are in wt%. If 60 lb oil are fed to the reactor per mol of gas, calculate the product oil composition. On Your Own Perform a DF analysis for this system. Click here to check your DF analysis of the oil desulfurization Now complete all the material balances. Click here to check your material balances for this process. 4.4.4 Formaldehyde Synthesis Problem Statement Formaldehyde can be produced by partial oxidation of methanol over a silver catalyst. Formic acid, hydrogen, carbon monoxide, carbon dioxide and water are also produced. An overall yield of formaldehyde of 2/3 is achieved in the process shown below. The reactor product is scrubbed to separate the light gases, and the residue is distilled to produce pure methanol for recycle.The scrubber and distillation units are combined into a single separator in the flowsheet. The reactor inlet stream is 35 mol% methanol and the off-gas composition is H2: 20.2%, CO2:4.8%, CO:0.2%, N2:74.5%, O2:0.3%. Calculate the compositions of all streams. On Your Own Perform a DF analysis for this flowsheet. Click here to see the DF analysis for the formaldehyde process. Complete the material balances for the formaldehyde process. Click here to view the material balances for this process. 4.5 Formal Stoichiometries This section deals with problems in which species balances are used in preference to element balances even if no reactions are given. We have two reasons for preferring species balances to element balances. There are always as many or more species balances as there are element balances. Also, the species balances are usually algebraically simpler. If the species present in the system are known, then the atom matrix for the system can be constructed. The technique of generating a feasible stoichiometric matrix from the atom matrix, and thus an independent set of reactions, was discussed in section 4.1.4. The problems in this section illustrate the use of this technique. 4.5.1 Ammonia Oxidation 4.5.2 Alcohols from Syngas 4.5.3 Formaldehyde Synthesis 4.5.1 Oxidation of Ammonia Problem Statement A feed stream formed from air and ammonia (20 mol air/mol NH3) reacts to produce a complex mixture consisting of H2O, O2, N2, NH3, NO, and NO2. When an 80% conversion of ammonia is achieved, the product stream contains 3% O2 and 6% NO. Calculate the compositions of the other species. On Your Own Perform a DF analysis on the ammonia oxidation reactor. Click here to check your DF analysis. Continuing with your worksheet, do the material balances. Click here to check your material balances. 4.5.2 Alcohols from Syngas Problem Statement When a mixture of CO and H2 (syngas) with a H2:CO ratio of 3.2 is reacted over a nickel catalyst, a mixture is produced that consists of hydrogen, carbon monoxide, water, methane, ethane, propane, methanol, ethanol, and carbon dioxide. When the conversion of CO is 97.9% and that of hydrogen is 73.8%, the product gas is characterized as follows: CO2/CO = 2.0, C2H5OH/CH3OH = 0.05, C3H8/CH4 = 0.01, C2H6/CH4 = 0.05. Find the composition of the product gas using only element balances. Repeat using only species balances. On Your Own Perform a DF analysis for this process. Click here to see a DF analysis of the alcohols from syngas process. In your worksheet, solve for the composition of the product gas using only the element balances. Click here to see the element balances for this process. In your worksheet, solve for the composition of the product gas using only species balances. Click here to see the species balances for this process. 4.5.3 Formaldehyde Synthesis Problem Statement Formaldehyde can be produced by partial oxidation of methanol over a silver catalyst. Formic acid, hydrogen, carbon monoxide, carbon dioxide and water are also produced. An overall yield of formaldehyde of 2/3 is achieved in the process shown below. The reactor product is scrubbed to separate the light gases, and the residue is distilled to produce pure methanol for recycle. The scrubber and distillation units are combined into a single separator in the flowsheet. The reactor inlet stream is 35 mol% methanol and the off-gas composition is H2: 20.2%, CO2:4.8%, CO:0.2%, N2:74.5%, O2:0.3%. Calculate the compositions of all streams using species balances. On Your Own Perform a DF analysis for this flowsheet. Click here to check your DF analysis for the formaldehyde process. Complete the material balances for the formaldehyde process. Click here to view the material balances for this process. 5.6.2 Multiple Non-Linear Equations 5.6.1 Numerical Techniques for Single Equations Solving Non-Linear Equations5.6 Acid-Gas Removal5.5 Methyl Iodide Synthesis5.4 Perchloric Acid Synthesis5.3 Simple Recycle5.2 Extending the Sequential-Unit Approach5.1 We follow the pattern established in previous chapters. The first section defines relevant terms and explores the special issues associated with this new technique. Subsequent sections apply the technique to selected process flowsheets. In all the flowsheets considered thus far we have implemented a sequential-unit solution strategy. We were able to solve all the balances for a selected process unit, then we moved on to the next unit. This chapter deals with flowsheets for which this procedure breaks down. Rather than abandoning the sequential-unit approach, we will augment our strategy by proceeding with partial solution of one or more sets of balances. Chapter 5 Sequencing with Partial Solution 5.1 Extending the Sequential-Unit Approach For all the flowsheets considered thus far, our calculation sequencing procedure has resulted in a partitioning of the material balances into ordered subsets that were completely and separately solvable. We simply identified a process unit with a degree-of-freedom of zero, solved the balances for that unit, and then used the new information thus obtained to reduce the DF for subsequent units to zero. This process was repeated until all the streams were fully characterized. Sequencing with Partial Solution This chapter deals with flowsheets whose specifications do not permit the balances for some unit(s) to be solved completely. Rather than abandoning the sequential-unit approach, we augment it with the following technique. 1. We first solve the balances for all units that have a zero DF, if there are any. 2. Then we consider the balances for that unit with the lowest DF. For this discussion, let the DF for this unit be '1'. The positive value of DF means that we will not be able to solve all the balances completely. Instead, we will have to estimate the value of one carry-forward variable, and express the solutions for all other computed quantities in terms of this variable. 3. We then continue with the balances for the remaining process units. 4. Eventually, since the DF for the process is zero, we will have a redundant balance that can serve as an "objective function". We can vary the value of the estimated variable in order to satisfy this objective function. Before giving more details about implementing the procedure of sequencing with partial solution, let us deal with some preliminary issues that sometimes cause confusion. An Alternative Strategy Instead of solving the material balances sequentially, unit by unit, we could write down all the balance equations at one time and solve them simultaneously. The fact that the DF for the process is zero means that this procedure can be made to work. Modern calculators and tools like Mathcad make the computational burden a manageable one. However, there are some good reasons why this is not a good approach for hand calculations. 1. There is the practical problem of selecting an independent set of balances. Verifying that a set of three or four balances for one unit is an independent set is much easier than trying to make this determination for a much larger set of equations. 2. It is easier to locate an error in one equation (or in a small set of equations) than it is to locate such an error in a large set of equations. A complex flowsheet with a dozen process units and several species may involve hundreds of variables. The sequential-unit approach is predicated on the principle that a small set of equations is easier for a human being to deal with than a large set of equations. This ability to "debug" a solution is a compellingreason to favor the sequential-unit strategy over the simultaneous-equation approach. 3. By dealing with the balances on a unit-by-unit basis, intuitive approaches for a particular application are more likely to be employed. These "obvious simplifications" are more likely to be recognized in the sequential approach than in the simultaneous-equation approach. Although it would be nice to believe that a solution to the general material balance problem is available in some sense, it is still true that leaps of intuition can often arrive at a solution with far greater efficiency than a "general" method. Although we are trying to develop a framework of tools to lessen the need for such intuitive approaches, many flowsheets display unique features which can be profitably exploited by an engineer with the skill to recognize them. Hand Calculation The reader may have wondered why we continue to employ he term "hand calculation" when we are clearly gearing our solution methods for implementation by computing tools such as Mathcad, or similar packages. We define a hand calculation as one in which we formulate the balances ourselves. Whether we use an abacus, slide rule, calculator, or Mathcad to actually solve those balances is irrelevant. These tools only handle the tedious mathematical manipulations required to solve the equations. We still have to tell the tool what equations to solve. There are software tools called process simulators which (with some restrictions on the specifications) could formulate as well as solve all the balances for us. All the user has to do is enter the data. Such a tool permits fully automated solution of all the material balances. This electronic book does not concern itself with such tools. We return now to the issue of determining the value of a carry-forward variable. Non-Linear Equations The final determination of the estimated variable is made by satisfying some objective function. The objective function is usually one of the redundant balances that involves the process unit for which the stream variable estimate was made. In one sense, the problem has been reduced to solving f(x) = 0, where f(x) is some non-linear expression in x. Evaluating the Objective Function Unlike the situation in which we have a single, equation to "plug into", it is usually the case that evaluating the objective function may involve performing many intermediate balances. In the figure below, let x be the carry-forward variable. There is no distinction in a mathematical sense between a closed-form expression, like f(x) = 3 x2 - sin(x) + 5, and a procedure that can uniquely assign a value to our objective function, given a value of the variable x. All that matters is that we can indeed evaluate the function. Converging to a Solution Once we have estimated the value of the carry-forward variable, say x, and selected an objective function f(x), we must determine the value of x that satisfies f(x) = 0. Distinguish clearly the process of evaluating the objective function and the process of solving the objective function. One way to solve the equation is to simply guess values of x until we get lucky enough to find one that makes f(x) =0. This method is called solving by "trial and error". It is usually impractical to proceed by this fashion. A better method is to guess only once, and then use an updating rule to successively refine the value of x, that is, converge to a solution. Many such updating rules exist. The Newton, Secant and Wegstein methods are examples of such convergence techniques. They all employ some rule for updating the current estimate of the solution. The Secant method is illustrated below. Other iterative methods are discussed in section 5.6. The sample flowsheet analyses in sections 5.2 - 5.5 illustrate the technique with far greater clarity than any word description can afford. The essence of the technique of carry-forward variables is simple. When the DF for a unit is positive, introduce one or more carry-forward variables. Values for these variables must be estimated. To determine the actual values of the variables, complete sufficient additional balances to permit evaluation of some objective function. If the objective function is not satisfied within some tolerance, then vary the assumed values of the variables by some convergence technique, and re-evaluate the objective function. Repeat this procedure until the objective function is satisfied to within the desired tolerance. Summary Err 6.381− 10 10−×=Err x6 2 1 3 −:= The last value of x is very close to the exact solution. x 1 1.5 1.21053 1.25141 1.26027 1.25992 1.25992 = The sequence of values generated is: xk 1+ xk 1− f xk( )⋅ xk f xk 1−( )⋅− f xk( ) f xk 1−( )−:=k 1 5..:= Currently, we have two argument values x1 = 1, x2 = 1.5 and two function values: f1 = -1, f2 = 1.375. Let us generate some new estimates for x. We will use index k to keep track of them. The updating rule for the secant method is f x1( ) 1.375=x1 1.5:= f x0( ) 1−=x0 1:= f x( ) x3 2−:= To illustrate the technique, let us solve f(x) = x3 - 2 = 0. The answer is the cube root of two. But let us converge to that value by starting with an initial estimate of 1.0. Secant Method 5.2 Simple Recycle Problem Description A feed of 1000 mol/h consisting of 1/3 A, 2/3 B is mixed with a recycle stream and reacts with the stoichiometry A + B --> D. The conversion of A per pass is 20%. The split fractions from the separator feed to the recycle stream are A:0.80, B:0.90, D:0.10. Calculate the flow rates of all streams. Click on the diamond icon to copy the figure and the problem statement to your worksheet. On Your Own Show that the problem is specified correctly and determine a calculation order. If any carry-forward variables are required, be sure to note when they will be needed, and how they will be determined. Click on this text to view the DF analysis for this problem. Set up and solve all the material balances. Click on this text to view the material balances for this problem. 5.3 Perchloric Acid Synthesis Problem Description Perchloric acid is produced by reacting barium chlorate with sulfuric acid. Ba(ClO4)2 + H2SO4 = BaSO4 + 2 HClO4 In the process shown below, an 80% conversion of barium chlorate is achieved when the ratio of sulfuric acid to barium chlorate in the combined reactor feed is 1:1.2. The feed (stream 1) contains 90 wt% barium chlorate in perchloric acid. The recycle stream contains only Ba(ClO4)2. Calculate all flowrates. Click on the diamond icon to copy the figure and problem statement to your worksheet. On Your Own Show that the problem is specified correctly and determine a calculation order. Note if any carry-forward variables are required. Click on this text to view the DF analysis for this problem. In your own worksheet, set up and solve all the material balances for this problem. Click on this text to check your material balances. 5.4 Methyl Iodide Synthesis Problem Description Two thousand lb/day of HI is added to an excess of methanol in the reactor shown below. The compositions in the product stream are given as mass fractions. The recycle stream is pure HI. What is the recycle ratio (stream 5: stream 1) if a single-pass conversion of 30% is achieved via the reaction HI + CH3OH = CH3I + H2O Click on the diamond icon to copy the problem statement and the figure to your worksheet. On Your Own Determine if the problem is specified correctly. Specify a calculation sequence, and determine if anycarry-forward variables are required. Click on this text to see the DF analysis for this problem. Solve for the required recycle ratio. Click on this text to see the material balances for this problem. 5.5 Acid-Gas Removal The absorber-stripper system shown in the figure removes CO2 and H2S from a feed consisting of 30% CO2 and 10% H2S in an inert gas. A proprietary solvent is used to selectively absorb the acid gases. The absorber bottoms are flashed. The ratio of the concentration of CO2 in the exit vapor (stream 6) to that in the liquid leaving the flash unit is 13.33. A similar ratio of 6.9 is measured for H2S. The liquid leaving the flash contains 5% CO2, and the overhead contains 20% solvent. The stripper overhead contains 30% solvent. The bottoms stream leaving the stripper is pure solvent. The feed is cleaned to the extent that stream 2 contains only 1% CO2 and no H2S. The splitter beneath the flash unit operates to produce equal flowrates in streams 3 and 8. Compute the compositions of all streams and the amount of solvent per mole of feed required. Problem Description On Your Own Is there sufficient data supplied to solve this problem? Click on this text to see the DF analysis for this problem. On your own worksheet, determine the flowrates of all streams. Click on this text to view the material balances for this problem. 5.6 Solving Non-Linear Equations Whenever a carry-forward variable, or tear variable, is required to proceed with the balances for a flowsheet, some technique must be implemented to converge on the correct value of the assumed quantity. If only a single tear variable is needed, then there are several techniques which can reliably converge to a solution. The Newton, Secant, Successive Substitution, and Wegstein methods are discussed in the first section below. The Wegstein convergence method is easily extended to deal with the case of multiple tear variables. This technique is discussed in section 5.6.2. 5.6.1 Numerical Techniques for Single Equations 5.6.2 Multiple Non-Linear Equations 5.6.1 Numerical Techniques for Single Equations The techniques in this section involve a very general class of equations. Any problem that can be phrased in the form f(x) = 0 can be treated by the methods discussed here. Whether a numerical method actually converges to a root, of course, depends heavily on the nature of the function whose root is sought. Closed-Form Objective Function We first consider the case when the function f(x) can be written in closed form. Such functions arise frequently when physical systems are simulated. An Example Suppose we are asked to find the value of the argument x when the functions y(x) = exp(x) and z(x) = 5 - 4(x-3/2)2 intersect? First, let's plot these functions to see what they look like. M 30:= i 0 M..:= Xi 2 i M ⋅:= y x( ) exp x( ):= z x( ) 5 4 x 3 2 − 2 ⋅−:= 0 0.5 1 1.5 2 0 2 4 6 y Xi( ) z Xi( ) Xi The function z(x) is an inverted parabola centered on x = 1.5. There are two roots, i.e. there are two values of x that satisfy the equation y(x) = z(x). These two roots lie near x = 0.6 and x = 1.6. General Form The problem can be restated in general as f(x) = 0, in which the function is f x( ) exp x( ) 4 x 3 2 − 2 ⋅+ 5−:= 0 1 2 2 0 2 f Xi( ) 0 Xi The function f(x) crosses the axis near x = 0.6 and x = 1.6. Suppose we wish to find the exact value of the root near x = 1.6? We already have a good estimate for the root. What we need is a technique to update and improve this estimate. Below, we develop such a technique. Later we will come back to this example and find the desired root. Newton's Method Illustrated above is an arbitrary function f of a single variable x. We seek to find the root x* given the estimate of the root: x0. Newton first suggested that an improved estimate of the root could be found by following the tangent to the curve back to where it meets the axis at x1. Two expressions can be written for the slope of the tangent: one is the derivative f'(x), the other is the divided difference representing the slope of the curve at x = x0. f' x0( ) f x0( ) 0−x1 x0−= This expression can be solved for the new estimate. x1 x0 f x0( ) f' x0( )−= Note that both the function f and its derivative f' are evaluated at x0. That is the entire right-hand side is a function only of x0. The calculated value x1 will in general be closer to x* than the starting value x0. The updating expression, or iteration formula can be applied multiple times to generate the sequence x2, x3, x4, .... which under certain conditions will converge as closely as desired to the root. That is, after we have calculated x1, we can use x1 on the right to generate x2, etc. We stop when the new value of x differs from the last value by some desired tolerance. Iteration Scheme Characteristics All iteration methods have the following characteristics: 0. An initial estimate must be supplied. This estimate is usually based on intuition about a physical system, or analysis of the characteristics of the mathematical expression. This is the place where "engineering problem solving" skills are applied. 1. An iteration formula must be available to generate new values of the argument. Newton's formula is an example. We will see others shortly. 2. Some convergence criterion must be selected. This is also an issue which depends on the nature of the analysis being carried out. In essence one must decide how many places of accuracy are reasonable for the argument of interest. xr 1.601209=xr root f ξ( ) ξ,( ):=ξ 1.6:= Of course, we could just as easily have let Mathcad solve for this root directly: f x3( ) 0=... and it is. It is clear that the third and fourth values are identical to six places. As a check, the function value should be near zero ... x 1.6 1.601211 1.601209 1.601209 =The sequence of values is: xk 1+ xk f xk( ) fp xk( )−:= ... by applying Newton's iteration formula: k 0 2..:=Generate a sequence of values ... x0 1.6:=The initial estimate of the root: fp x( ) exp x( ) 8 x 3 2 − ⋅+:= We need the derivative of f: f x( ) exp x( ) 4 x 3 2 − 2 ⋅+ 5−:= Newton's formula is applied below to the problem of finding the root of f(x) near x = 1.6. Newton Method Example Extension to Implicitly-Defined Functions Whenever an analytical expression for a function is available, it is clear that the Mathcad root function will solve for the desired root. However, Mathcad cannot deal directly with an implicitly defined function such as those which arise in the analysis of process flowsheets. The "function" can be visualized in the following context: The argument "x" is some flowsheet variable, such as a species composition, flow rate, or a dimensionless equivalent of such a quantity. The objective function is some material balance, an expression involving other flowsheet variables. Values for these other flowsheet variables are computed from other balances, independent of the balance that is used as the objective function. The values of the compositions and flow rates that appear in the objective function depend on the value of the argument, even if the argument itself does not appear explicitly in this balance. A critical distinction must be made between the process of evaluating this function and the method used to solve it, that is, to find the value of the argument that satisfies the balance. The evaluation consists simply in performingall the necessary intermediate balances. The "solving" consists in the application of an iterative scheme that converges to the desired root. The Newton method, which requires an explicit derivative, is not suitable for this task. Instead, a related method in which the derivative is estimated numerically can be employed. This method is termed the secant method. x1 x0 0.05+:=Another value of the argument: x0 1.6:=Initial estimate of the root: f x( ) exp x( ) 4 x 3 2 − 2 ⋅+ 5−:= We can apply this technique to the function defined above.Secant Application This is the secant updating rule. x2 x0 f x1( )⋅ x1 f x0( )⋅− f x1( ) f x0( )−= This expression can then be solved for the next value of x in the sequence. f x1( ) f x0( )− x1 x0− f x1( ) 0− x2 x1− = We equate two expressions for the slope of the secant at x1. The Newton method was based on following the tangent to the curve at x0 back to the x-axis. The tangent can be approximated by the secant to the curve through two closely-spaced points, say x0 and x1. Secant Method k 0 7..:=x0 0.5:=Initial estimate: g x( ) 1 2 x2⋅+ x2.2− 2.8 :=One reformulation is f x( ) x2.2 2 x2⋅− 2.8 x⋅+ 1−:= A simpler technique for solving f(x) = 0 is to reformulate the objective function (when possible) in the form x = g(x). The right-hand-side is then an updating rule. For the function Successive Substitution If the convergence tolerance δ = 0.001 then we could have stopped with the fourth value (the second iteration) in the sequence. With δ = 0.000001 we needed six values (four iterations). xk xk 1−− δ< In general, iteration is stopped when a convergence criterion like x 1.6 1.65 1.6011462 1.6012062 1.6012095 1.6012095 = The method converged to the same root as before, albeit more slowly. xk 2+ xk f xk 1+( )⋅ xk 1+ f xk( )⋅− f xk 1+( ) f xk( )−:= k 0 3..:=Let's generate a sequence of values ... Generate the sequence: xk 1+ g xk( ):= The sequence converges slowly. x 0.5 0.45799 0.44289 0.43773 0.43599 0.43542 0.43523 0.43516 0.43514 = Convergence Problem Although successive substitution is temptingly simple, it often fails to converge. We can try solving the function used in the Newton and Secant examples by this technique to see this behavior. f x( ) exp x( ) 4 x 3 2 − 2 ⋅+ 5−:= g x( ) 5 exp x( )− 4 3 2 +:= Initial estimate: x0 1.601:= Generate the sequence: xk 1+ g xk( ):= The sequence fails to converge even when supplied with an initial estimate that is very close to the root. x 1.601 1.602 1.593 1.642 1.5 0.205i+ 1.962 0.246i− 1.79 0.748i+ 2.269 0.662i− 2.196 1.068i+ = x 1.601 1.6024842 1.5930696 1.5997952 1.6013528 1.6011789 1.6012104 1.6012095 1.6012095 1.6012095 =The sequence converges. Sk 1+ xk 2+ g xk 1+( ) g xk( )− xk 1+ xk− xk 1+ 1 θ Sk( )−( )⋅ θ Sk( ) g xk 1+( )⋅+ := Generate a sequence: S0 x1 0 g x0( ) :=k 0 7..:=x0 1.601:=Initialization: g x( ) 5 exp x( )− 4 3 2 +:=Auxiliary functions: θ η( ) 1 1 η− := f x( ) exp x( ) 4 x 3 2 − 2 ⋅+ 5−:=We apply the Wegstein method to:Example 2. Update: x = (1 - θ) x1 + θ g(x1) 1. Evaluate x1 = g(x0), g(x1), S = ∆g/∆x, and θ = 1/(1-S) 0. Estimate the argument value x0, and supply a tolerance δ. This step-limited version of successive substitution displays better convergence properties. This technique also requires that the objective function f(x) = 0 be reformuled as x = g(x). The algorithm is: Wegstein Iteration Back to Reality The secant method is geared to solving f(x) = 0 while the successive substitution and Wegstein techniques are applied to problems that can be phrased as x = g(x). Let's recall why we are interested in solving such equations. In the simple recycle system shown below, let us assume that the DF analysis indicates that a carry-forward variable in stream 2 is required to initiate the material balances on the reactor. Let this variable be the flow rate of species A. That is, the argument of our objective function is x = nA2. We proceed by performing the balances on the reactor, the separator, and the overall subunit. The mixer balance is our objective function. Secant Formulation The objective function, the mixer balance on A, can be written as an error function: f x( ) nA1 nA5+ nA2−= 0= This formulation is tailor-made for the secant algorithm. Wegstein In this case we formulate the mixer balance as the function g(x): g x( ) nA1 nA5+= Summary Whenever a tear variable is required to proceed with the material balances in a process flowsheeet, it is necessary to implement some technique for converging to the correct value of the assumed variable. In essence, the problem has been reduced to the solution of a single equation in a single unknown. The Newton, Secant, Successive Substitution, and Wegstein techniques were presented for the case of an analytic function of a single variable. The only difficulty in applying these ideas to the solution of the material balances for a process flowsheet is that the evaluation of the objective function may involve the evaluation of many intermediate balances. y0 0.5:=x0 1:=Initial estimate of the roots: g2 x y,( ) 2 x 7 y⋅+ 1 2 +:=y g2 x y,( )= g1 x y,( ) x2 5 y 10 + 0.8+:=x g1 x y,( )= As was the case for a single equation, the system must first be reformulated for updating. One such reformulation is: Multidimensional Successive Substitution The desired root is at (1.139,0.798). f2 x y,( ) 2 x 7 y⋅+ 1 2 + y−:= f1 x y,( ) x2 5 y 10 + 0.8+ x−:= The system to be considered consists of two non-linear equations in two unknowns: x, y. Sample System All the methods discussed in the previous section can be extended to the case of finding roots to multiple, non-linear equations. However, these methods differ markedly in the relative amount of computation involved in the iterative algorithm. Each iteration may involve multiple evaluations of the objective functions. Due to the implicit nature of the objective functions involved in flowsheet analysis, and the subsequently heavy computational load involved simply in function evaluation, only the two simplest algorithms are discussed below. These are Successive Substitution and the Wegstein method. If the degree-of-freedom analysis for a given flowsheet indicates that more than one tear variable is required to proceed with the material balances then one is faced with the task of converging simultaneously on two or more arguments. 5.6.2 Multiple Non-Linear Equations Sxk 1+ xk 2+ Syk 1+ yk 2+ g1 xk 1+ yk,( ) g1 xk yk,( )− xk 1+ xk− 1 θ Sxk( )−( ) xk 1+⋅ θ Sxk( ) g1 xk 1+ yk 1+,( )⋅+ g2 xk 1+ yk 1+,( ) g2 xk 1+ yk,( )− yk 1+ yk− 1 θ Syk( )−( ) yk 1+⋅ θ Syk( ) g2 xk 1+ yk 1+,( )⋅+ := Iterate: θ s( ) 1 1 s− :=Auxiliary function: Sx0 x1 Sy0 y1 0 g1 x0 y0,( ) 0 g2 x0 y0,( ) := y0 0.5:=x0 1:=Initialize: This algorithm is applied to the same system used to illustrate the multidimensional successive substitution method. 0. Estimate the array of argument values x0 1. Calculate x1 = g(x0), the slopes Si = ∆gi/∆xi, and θi = 1/(1-Si) 2. Update with x2 = (1 - θ) x1 + θ g(x1) The algorithm is similar to that used in the one-dimensional case: Multidimensional Wegstein Iteration y 0.5 0.944 0.761 0.81 0.794 0.799 0.797 =x 1 1.05 1.115 1.125 1.134 1.137 1.138 = The sequence converges to the desired multidimensional root. xk 1+ yk 1+ g1 xk yk,( ) g2 xk yk,( ) :=k 0 5..:=Generate the sequence: The convergence is more rapid in this case. x 1 1.05 1.115 1.132 1.139 1.139 1.139 1.139 = y 0.5 0.944 0.761 0.796 0.798 0.798 0.798 0.798 = Relation to Flowsheet Analysis Let us recall once more the context in which multiple non-linear equations arise in process flowsheet analysis. It may occur that a degree-of-freedom analysis indicates that several tear variables may be required to initiate balances. For example, assume that the flow rates of all species in stream two below must be estimated in order to proceed with the balances on the reactor. The unknowns then are the species flow rates in stream two ni2. One possible scenario is that the reactor, separator, and overall subunit balances can then be completed, leaving the mixer balances as an objective function. The formulation for S species is: ni2 ni 1, ni 5,+= i 1 S..:= This form is precisely that needed to implement successive substitution or Wegstein iteration x = g(x). Summary The extension of the successive substitution and Wegstein methods to multiple non-linear equations involves no new principles. The convergence of systems of more than two equations is much slower than for the case of a single equation.
Compartilhar