An Acetaldehyde Supply Mechanism for the Improved Production of Pentaerythritol
COPYRIGHT AND CITATION CONSIDERATIONS FOR THIS THESIS/ DISSERTATION
o Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
o NonCommercial — You may not use the material for commercial purposes.
o ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.
How to cite this thesis
Surname, Initial(s). (2012) Title of the thesis or dissertation. PhD. (Chemistry)/ M.Sc. (Physics)/ M.A. (Philosophy)/M.Com. (Finance) etc. [Unpublished]: University of Johannesburg. Retrieved from: https://ujdigispace.uj.ac.za (Accessed: Date). AN ACETALDEHYDE SUPPLY MECHANISM FOR THE IMPROVED PRODUCTION OF PENTAERYTHRITOL
by
Graham Robert Jennery
Submitted to Technikon Witwatersrand in part fUlfilment of the requirements for the Masters Diploma in Chemical Engineering.
october 1992
supervisor: Dr. K. Richter
Co-supervisor Mr. L.A. Van Niel To Susan Thanks for so much.
To Eileen May you reach your goals. i
SUMMARY
The work presented here constitutes an account of the optimization of a chemical reaction process with special reference to the methodology of reagent addition in the case of fast reactions.
The chemical reaction process for manufacture of the Formaldehyde Acetaldehyde condensation product Pentaerythritol (Penta) as it is conducted at the plant of National Chemical Products, a division of Sentrachem, was studied in detail. The industrial scale reactor design was critically examined, 'with emphasis on the evaluation of mixing and reagent dispersion efficacy and its effect on chemical reactor performance.
Batches of Pentaerythritol products were prepared in a laboratory bench scale reactor. Reagent concentrations and proportions were controlled· at various values and the reaction temperature profiles were controlled so as to be similar for all the tests. Moreover the mode and intensity of liquid agitation and reagent admixture was varied in a controlled manner between the various tests.
The reaction liquors from the various batches were sampled and the samples sUbjected to chemical analysis. The results were then compared in order to show the effect of agitation and reagent dispersion intensity on the reaction process and products.
The results indicate conclusively that liquid flow or agitation intensity and reagent admixture or sparging variation has an effect on the type and relative amounts of chemical species produced in the laboratory apparatus. ii This effect is especially significant with respect to the side products Di-pentaerythritol and Bis Penta Mono-formal (B.P.M.F.). The effect is also demonstrated for the gamut of various side products collectively and arbitrarily designated as the so-called "unknowns". Furthermore the formation of coloured products in the reaction· is distinctly influenced by the same variation.
High intensity agitation and reagent sparging enhances Di penta formation and inhibits formation of the Formal B.P.M.F, "unknowns" and colour. At very low agitation and sparging intensity Di-penta production is diminished while B.P.M.F., "unknowns" and colour formation is favoured.
The work includes a proposal for the manufacture and installation of improved reagent sparging systems in the NCP Transvaal commercial scale Penta Reactor.A tentative method for the design of a continuous reactor for penta production using optimised Tee mixers for high velocity in-line reagent sparging is also developed. iii
ACKNOWLEDGEMENTS
Various people have in many ways, knowingly and otherwise contributed towards the present work. This has given me the opportunity to study an interesting but often little understood chemical reaction engineering topic, which has led to new light being shed on a special aspect, which is rewarding in itself. The help of the following persons is gratefully acknowledged
L.A. Van Niel of NCP Projects and Engineering Division for his suggestion of the topic in the first instance, support generally, and constructive criticism;
D.P. Bowyer, General Manager of NCP Transvaal, and to NCP management for permission to present this work as a dissertation to Technikon Witwatersrand;
Dr. K. Richter of Technikon Witwatersrand for his supervision, constructive criticism and encouragement;
M.A. McEvoy of NCP Transvaal for support, suggestions, ideas and many hours of discussion ;
Dr. H.A.H. Laue, Dr. C. D. simon, and members of the NCP Research & Development team for advice and suggestions, support and technical assistance ;
C. Manushewitz, NCP Projects and Engineering Division, for helpful practical discussion of reactor details ;
J.L. Steyn, P.F. Tack and the Pentaerithritol and Penta chemicals production and technical team members at NCP for support and assistance; iv
Miss T. C. Mvelase of NCP R&D Process Chemistry, who painstakingly provided expert technical assistance with the laboratory testwork ;
Miss H. Comninos of NCP R&D Analytical Chemistry, who diligently performed the numerous specialised chemical analyses ;
E. W. James of NCP Projects and Engineering Division, who prepared the figures in Chapters 4 and 8 using a PC CAD package. v
TABLE OF CONTENTS PAGE.NO.
SUMMARY i
ACKNOWLEDGEMENTS iii
LIST OF FIGURES viii
NOMENCLATURE x
CHAPTER 1 : INTRODUCTION AND OBJECTIVES 1
1.1 INTRODUCTION: HISTORICAL AND GENERAL 1
1.2 THE NCP PROCESS AND DISADVANTAGES 3
1.3 AIMS AND OBJECTIVES 4
CHAPTER 2 : PENTAERYTHRITOL PROCESS CHEMISTRY 6
2.1 PREPARATION OF PENTAERYTHRITOL INTRODUCTION 6
2.2 PENTA PREPARATION CHEMISTRY: A DETAILED DESCRIPTION 8
2.2.1 ALDOL CONDENSATION 8 2.2.1.1 Mechanism 8 2.2.1.2 Kinetics 9 2.2.1.3 Conclusions 11 vi
2.2.2 CANNIZZARO REACTION 13 2.2.2.1 Mechanism 13 2.2.2.2 Kinetics 14 2.2.2.3 Conclusions 14
2.2.3 PENTAERYTHRITOL FORMATION 15 2.2.3.1 The specific reactions 15 2.2.3.2 Side reactions 15
2.3 SUMMARISED CONCLUSIONS 17
2.4 RECENT RESEARCH SOME SPECIAL ASPECTS 18
CHAPTER 3 CHEMICAL REACTORS FOR PENTA PRODUCTION 20
3.1 CHEMICAL REACTORS AN INTRODUCTION 20
3.1.1 THE IDEAL BATCH REACTOR 20
3.1.2 STEADY-STATE MIXED FLOW REACTOR 21
3.1.3 STEADY-STATE PLUG FLOW REACTOR 22
3.1.4 IDEAL AND NON-IDEAL FLOW IN REACTORS 24
3.2 REACTORS FOR THE PREPARATION OF PENTAERYTHRITOL 25
3.3 SOME INDUSTRIAL PROCESSES 26
3.3.1 HERCULES POWDER CO. PROCESS 26
3.3.2 JOSEF MEISSNER PROCESS 28 3.3.2.1 Meissner process with 28 Sodium Hydroxide catalyst vii
3.3.2.2 Meissner process with 28 Calcium Hydroxide catalyst
3.3.3 PERSTORP PROCESS 31
CHAPTER 4 : THE NATIONAL CHEMICAL PRODUCTS PROCESS 32
4 . 1 THE CHEMI CAL REACTOR 32
4.1.1 NCP PRIMARY REACTOR AND COOLING CIRCUIT 32
4.1.2 REAGENT ADDITION AND SPARGING SYSTEM 36
4.2 REACTION PROCESS CONDITIONS 40
4.2.1 MOLE RATIO 40
4 . 2 . 2 REAGENT STRENGTHS 40
4.2.3 REACTION TEMPERATURE 40
4.2.4 REAGENT ADDITION PROCEDURE 41
CHAPTER 5 PROBLEM, STRATEGY AND EXPERIMENTAL DESIGN 42
5.1 THE NCP REACTOR PROBLEM 42
5.2 OBJECTIVE STATEMENT AND PRACTICAL CONSTRAINTS 43
5.3 EXPERIMENTAL DESIGN 44 viii
CHAPTER 6 THE EXPERIMENT 47
6.1 TEST MATERIALS 47
6.2 EXPERIMENTAL APPARATUS 48
6.3 EXPERIMENTAL PROCEDURE 52
6.3.1 ADJUSTMENT OF INDEPENDENT VARIABLES 53 6.3.1.1 Agitation and reagent dispersion 53 6.3.1.2 Mole ratio 53
6.3.2 REAGENT ADDITION RATE 54
6.4 CHEMICAL ANALYSES 54
6.5 CRITICAL EXAMINATION OF RESULTS 54
6.6 TESTWORK SETS 55
CHAPTER 7 RESULTS AND DISCUSSION 56
7.1 RESULTS 56
7.2 DISCUSSION 63
7.3 CONCLUSIONS 64
CHAPTER 8 : REAGENT SPARGING SYSTEMS 66
8.1 PENTA REACTOR SPARGING SYSTEMS: 66 PRACTICAL CONSIDERATIONS ix
8.2 THE DESIGN OF REAGENT SPARGING EQUIPMENT 68 FOR THE NCP REACTOR
8.2.1 ACETALDEHYDE SPARGER 68 8.2.1.1 Calculations: Acetaldehyde sparger 69
8.2.2 CAUSTIC SODA SOLUTION SPARGER 72 8.2.2.1 Calculations: Caustic sparger 73
8.3 ALTERNATIVE SPARGING METHODS 77
8.4 CONCLUSION: A RECOMMENDATION 78
FOOTNOTE TO CHAPTER 8 : SPARGERS : IMPLEMENTATION 79
CHAPTER 9 : TEE MIXERS 80
9.1 TEE MIXERS: GENERAL 80
9.2 A TEE MIXER FOR THE NCP REACTOR 81
CHAPTER 10 STEADY-STATE FLOW REACTORS 85
10.1 STEADY-STATE REACTORS GENERAL 85
10.2 REACTORS FOR PENTA: ESSENTIAL REQUIREMENTS 87
10.3 A CONTINUOUS PENTAERYTHRITOL REACTOR 87
10.3.1 CONCEPTUAL DESIGN 87
10.3.2 DISCUSSION 90 x
10.4 RECENT RESEARCH 91 AND INDICATIONS FOR THE FUTURE
APPENDIX 1 : REACTOR PARAMETERS 92
A1.1 DISCHARGE FLOW NUMER, Nq 92
A1.2 AGITATION INTENSITY NUMBER, N, 94
. A1. 3 BLEND TIME, 8n 94
A1.4 POWER NUMBER" Np 95
A1.5 REYNOLDS NUMBER, . Re tllll 95
APPENDIX 2 : CALCULATION OF RESULTS FROM EXPERIMENTAL DATA 96
A2.1 PRODUCTS AS % mjm OF TOTAL PRODUCTS 96
A2.2 YIELDS BASED ON ACETALDEHYDE 97
A2.3 SELECTIVITIES 98
A2.4 EXPERIMENTAL RESULTS 98
REFERENCES 102 xi
LIST OF FIGURES Figure page No.
2.1 Dependence of yields of pentaerythritol 18 on reaction starting conditions.
3.1 Classification of reactor types 23
3.2 Flowsheet of Hercules Powder Company Penta 27 process.
3.3 Flowsheet of Josef Meissner Company Penta 30 process
4.1 NCP Primary reactor system, cooling circuit and reagent charging. 35
4.2 NCP primary stirred tank reactor showing principal dimensions. 38
6.1 Laboratory bench scale reactor for pentaerythritol preparation. 50
6.2 Bench scale test rig for reactor parameter experiments with controlled stirring and reagent sparging. 51
6.3 Bench scale stirred tank reactor showing principal dimensions. 51
7.1 Monopenta as % mjm of total products. 58
7.2 Penta by-products as % mjm of total products. 58 xii
7.3 Pentaerythritols ( Mono + Di ) as % m/m of total products. 59
7.4 Yields based on Acetaldehyde. Mono-Penta : Moles per mole. 59
7.5 Yields based on Acetaldehyde. Di-Penta and B.P.M.F. : Moles per mole. 60
7.6 Molar Selectivity for Mono-penta. 60
7.7 Molar Selectivity for Di-penta. 61
7.8 Range of results Mono -penta as % m/m of total products. 61
7.9 Range of results Di-penta as % m/m of total products. 62
7.10 Range of results: B.P.M.F. and "Unknowns" as % m/m of total products. 62
8.1 Stirred tank reactor showing present reagent "Dip tubes". 74
8.2 Stirred tank reactor showing proposed ring spargers. 75
9.1 90° Tee Mixer. 84
10.1 n stirred tank reactor stages in series. 89
10.2 "Side Feed Flow" steady-state reactor. 89 xiii
NOMENCLATURE symbol Dimensions
Cross sectional area of jet orifice
B Baffle width in agitated tank [m]
C Agitator turbine off-bottom distance [m]
C~ Initial or entering concentration of species A in a chemical reactor [mol cm.~]
CB Baffle distance in an agitated tank [m]
CD Frictional discharge coefficient for orifices, [Dimensionless]
Cs Interstagedistance in mUltistage agitated tank [m]
D Pipe diameter [m]
DA Agitator impeller diameter [m]
D~ Dip pipe distance (radial) [m]
D~ Sha f t diameter [m]
D~ Sparger ring diameter [m]
Dr Tank diameter [m]
d Diameter of jet or side tee
do Diameter of sparger hole [m] xiv
FAO Entering flow rate of species' A in a chemical reactor [mol s·11
I G Mass flow rate [kgs· ] g Acceleration due to gravity = 9.8066
Hdl' Dip tube height (of holes) [m ]
HL Liquid height in an agitated tank [m]
h o Orifice pressure loss as head [m]
L Agitator Turbine blade length [m]
1 Impingement distance [m]
N Shaft speed
NAO Initial or entering amount of species A in a chemical reactor [moles]
N, Agitation Intensity number [Dimensiomless]
NI' Power number of agitation [Dimensionless]
Nq Discharge flow number [Dimensionless]
P Power drawn in agitation system [W]
Q circulation flow produced by agitation impeller q Fluid flow rate in a pipe
q~d Discharge flow produced by an impeller in a radial 3s·1 direction [m ] xv
Re Reynolds number [Dimensionless]
Reo Orifice Reynolds number [Dimensionless]
Rem Reynolds number of mixing [Dimensionless]
r A Rate of reaction or formation of species A based on volume of fluid [moles em" S-I] s Space velocity [S-I]
TA Impeller shaft Torque in agitation system [Nm] t Time [s] u Jet fluid velocity [rns"]
Uo Orifice fluid velocity [ms-I ]
3 V Fluid Volume [m ] v Pipeline fluid velocity [ms"]
j Vb Bulk fluid velocity [ms- ] v=x Maximum velocity at centreline of agitation impeller (either radial or resultant) [ma"] w Agitator Turbine blade width [mJ
XA Fraction of reactant A converted into product [Fraction] x Downstream distance from side tee [m] xvi
Z Number of stages in 'a mUltistage agitation system [Number] z Distance across pipe at side tee entry [m]
Greek symbols
~ A constant of limited value range, from a paper by Forney and Gray (1990). [Dimensionless] ea Blend time in mixing vessel [s] eTO Tank recirculation turnover time [s]
Liquid viscosity [Pas]
O p Liquid density [kg m "]
T = C/\ttV/FAtt , = (lIs) , Space-time i.e. time required to process one reactor volume of feed [s] 1
CHAPTER 1 INTRODUCTION AND OBJECTIVES
In this chapter the reader will be given a brief introduction to the chemistry and technology of the pentaerythritol manufacturing process. The specific application which is to be studied, i. e. the industrial scale production reactor used at National Chemical Products, and its associated disadvantages will also be discussed. Finally the overall objective of the work described in subsequent chapters and the way in which specific problems are to be solved will be given.
1.1 INTRODUCTION: HISTORICAL AND GENERAL
The polyhydric alcohol Pentaerythritol (Penta) was discovered quite accidentally, like many other important chemical compounds, by Tollens in the late 19th century. While investigating the reaction of formaldehyde with barium hydroxide he encountered small quantities of a white precipitate which later proved to be due to the presence of acetaldehyde as an impurity. A report on the formation of this product from acetaldehyde, formaldehyde and lime was published by Tollens and Wigand in 1891. Considerable interest developed in penta and its derivatives and commercial production started in the early 1930's. Several American companies began to produce the highly purified compound for use in the manufacture of the explosive PETN (pentaerythritol tetranitrate).
Penta is manufactured by an exothermic condensation reaction between Acetaldehyde and Formaldehyde, the latter in excess of stoichiometric quantity, with an alkaline catalyst such as caustic soda or lime. The chemical reactions take place in a single aqueous phase and can be carried out as a batch operation or as a continuous process although the former seems to be more commonly practised. 2 Various by-products and pentaerythritol polymers, mainly Di-Penta, sometimes Tri-Penta, BisPenta MonoFormal (B.P.M.F.) and some cyclic products, are also produced. On completion of the reaction excess alkali is usually neutralised, commonly with· Formic acid but it is also possible to use acetic, sulphuric or phosphoric acid. The penta products, as well as alkali formate· and excess Formaldehyde are recovered by filtration and drying operations and evaporation or distillation.
Pentas are important raw materials for the manufacture of various chemical commodities such as drying oils, plasticisers, polyesters, alkyd resins and other compounds used in surface coatings. Pure mono-pentaerythritol, known as "Nitration grade" is still used, as mentioned above, in the manufacture of high powered modern explosives. Miscellaneous uses include the manufacture of fire retardant coatings, heat stabilizers for polyvinyl chloride, and as a sensitizer, with ascorbic acid, in the catalytic oxidation of methyl orange and other dyes. A dry solid formulation prepared from ethylene glycol and a polypentaerythritol can be used as a friction dressing for power transmission belts.
The chemistry of Pentaerythritol formation and the effects of reaction conditions, e.g. reagent mole ratios and reaction temperature profiles have been studied and reported previously, mainly for laboratory scale batch reactors and very often confidentially by "in house" investigators, as for example the work done at NCP (JennerY,1991"i Laue,1984i Simon, 1991.i Stoltz & salemi, 1979b )• Some knowledge is in the pUblic domain, examples being the material given in the section "Polyhydric alcohols " in the Kirk Othmer Encyclopedia and the A.C.S. monograph (Berlow,Barth & Snow,1957). A considerable amount of information seems to have been generated by workers in Eastern block countries, much of 3
which has been publ Lshed in the Russian or other Slavic language journals e.g. the work of Belkin(1979) , and Koudelka (1984) . There are also various patents pertaining to commercial processes. A recent NCP R&D review by
Simon (1991h ) has surveyed the significant relevant literature available to date. However one aspect of reactor design which seems to have been neglected in all the previous literature on Penta formation is the effect of fluid flow and reagent sparging and dispersion parameters, a topic within the domain of chemical reaction engineering. This is surprising in view of the fact that the single aqueous phase chemical reactions under consideration are rather fast.
Which brings us to a discussion of the aims and objectives of the following chapters and more specifically an introduction to the laboratory experimentation programme.
1.2 THE NCP PROCESS AND DISADVANTAGES
The present industrial scale production unit at NCP Transvaal consists of a stirred tank batch reactor with a recirculating cooling circuit. It has long been felt by various persons that there were certain inadequacies in the mixing and reagent distribution systems of this equipment. L.A. van Niel of NCP Projects and Engineering Division has contended that previouslY,in an old reactor, the process had a better sparging system in that the reagents were introduced into the turbulent recirculation line. Drawings provided by C Manushewitz also of NCP PED show that Perstorp A.B., a major European firm uses equipment having sparging devices of refined design.
A mUlti-disciplinary NCP research team has been investigating various aspects of penta process chemistry and manufacturing technology. Major contributions included the development of accurate and reliable analytical methods 4 for determination of pentas and other species in reaction liquors and products. It was felt that this new capability would enhance the success of an effort directed at the investigation of reactor parameter effects such as reagent proportions and reaction temperatures. Some of these factors are presently being studied at plant and laboratory scale by team members. Mixing and reagent sparging parameter investigations, on the plant if possible, were desirable but due to insufficient knowledge generally, and the perceived risks associated with plant configuration changes, full scale experimentation could not be justified. It was felt that more specific information should be obtained before any experiments involving equipment modifications could be initiated.
1.3 AIMS AND OBJECTIVES
In the light of the foregoing it was decided to attempt laboratory bench scale experimentation to investigate the effect on the reaction of changes in fluid flow and reagent sparging characteristics. This was in spite of cautionary comments by van Niel(1991) and also by the present author (Jennery,1991.) pointing out that meaningful results from small scale testwork would be rather difficult to achieve. It was appreciated that measurements of fluid flow parameters in the laboratory glassware available would be very approximate and for this reason the experiments were conducted only at the extremes of flow conditions attainable. It was thus planned to show trends and approximations at best, and numerical results would not necessarily be expected to relate simply or directly to Industrial scale processes. Development of the experimental techniques constituted a major part of, and prelude to, this work.
In a later chapter it will be shown that notwithstanding these difficulties the experimentation was successful. The 5 results obtained clearly indicate that variation of agitation intensity and reagent sparging mode have an effect on the relative amounts of the different chemical species produced in the penta formation process. But first in the following pages a more detailed overview is given of the chemistry and of the technological aspects of pentaerythritol production. 6
CHAPTER 2 : PENTAERYTHRITOL PROCESS CHEMISTRY
The chemistry of pentaerythritol formation has been studied extensively by various investigators. A comprehensive review has been given by Laue(1984) of NCP Research and Development. The section 2.1 of this chapter is a general introduction to the topic. In section 2.2, following, and sub-sections a more detailed treatment is presented. The summing-up in section 2.3 represents an essential view of and generally accepted interpretations of the penta process chemistry knowledge at NCP to date. Some special relevant aspects which have been highlighted in recent research are shown in section 2.4.
2.1 PREPARATION OF PENTAERYTHRITOL INTRODUCTION
Penta is a polyhydric alcohol and the generally accepted preparation route is an aldol condensation between Acetaldehyde and Formaldehyde followed by a crossed Cannizzaro reaction with caustic soda as the catalyst.The chemistry can be simply represented as shown below :
CH2CHO + 4HCHO + NaOH = + NaHCOO
In general the Aldol Condensation occurs with Acetaldehyde in the presence of dilute sodium hydroxide,potassiurn carbonate or hydrochloric acid to form a syrupy liquid known as Aldol.
NaOH (50 %)
-+ 7
Aldol condensation can occur (1) between two aldehydes (identical or different); (2) between two ketones (identical or different); and (3) between an aldehyde and a ketone. Whatever the nature of the carbonyl compound , it is only the a-hydrogen atoms adjacent to the carbonyl group which are involved.
Aldehydes that have no a-hydrogen atoms undergo the Cannizzaro reaction in which two molecules of the aldehyde are involved, one being converted .into the corresponding alcohol and the other into the acid. The usual reagent for bringing about this reaction is 50 % aqueous or ethanolic alkali, e.g.,
2H.CHO + NaOH H. CO.,Na + CH~OH
Although this reaction is characteristic of aldehydes having no a-hydrogen atoms it is not restricted to them only.The Cannizzaro reaction can also take place between two different aldehydes, and is then known as a "crossed" Cannizzaro reaction. Formaldehyde can participate in such a "crossed " reaction, and the nature of the product depends on the structure of the other aldehyde.A special case is the reaction with acetaldehyde, which has 3 a-hydrogen atoms. Tetrakishydroxymethylmethane (tetramethylolmethane) or pentaerythritol is formed:
CH~.CHO + 4H.CHO NaOH C (CH~OH) ~ + H. CO.,Na
The reactions have been treated in greater detail in the survey compiled by Laue, of which some salient points are discussed in the following section. 8
2.2 PENTA PREPARATION CHEMISTRY A DETAILED DESCRIPTION
2.2.1 ALDOL CONDENSATION references: Gould (1959), Ingold (1959), March(1968).
2.2.1.1. Mechanism
This will be discussed specifically for the acetaldehyde/formaldehyde condensation.
The reaction can be represented as follows
o o o II II II II
( 1) CH3-CH + OR [ - CH2 -CH k., I
0 0 O' 0
II II II II II k 2 II (2) HCH CH., -CH ---- H2C -CH2 -CH k. 2
II
k,
III
Since incorporation of Deuterium onto the methyl group of acetaldehyde does not occur when the reaction is carried 9 formation of I irreversible. ( See kinetics however)
Similarly k) is very fast compared to k.2 , the reaction thus being essentially irreversible.
The reaction (2) of the carbanion with formaldehyde is favoured rather than reaction with another acetaldehyde molecule; firstly by the fact that the formaldehyde carbon has less electron density and is therefore more positive; secondly by keeping the acetaldehyde concentration low; thirdly by using excess of formaldehyde.
It is interesting to note that pentaerythritol has been prepared from the acetaldehyde condensation products, acetaldol and crotonaldehyde ( IV and V ).
CH~. CHOH. CH1 • CHO CHI. CH=CH • CHO IV V
The reaction mechanisms are given in reference 2.
The implication is that the reversibility of the condensation in the case of acetaldehyde/acetaldehyde must be substantial.
2.2.1.2. Kinetics
In a more generalised form the reactions discussed , are
4) B- + E - HB + E- k.1( HI!)
5) E. + CH20 ECH,O· 10
k~
6) ECH.,O - + H.,O ;::! ECH.,OH + OH -
k_~
where E = Enolisable aldehyde ( e.g. acetaldehyde) B = Base
Experimental observations show that the rate of dissappearance of E follows 1st order kinetics to at least 95% completion. Also, reactions go more or less to completion ( overall irreversibility ), and it is known that in aqueous solutions reaction (6) is very fast.
From this the following rate law is derived:
7) f!l.ll = - k 2.1.kl fOH-1 [OH' + ~IIJ5.lfII-1 [B-'} [ E) [ CH 2Ql
dt k-I(H2U) [H:P] + z k-I(HIII [HB ] + k.,[ CH.,O]
From 7) two limiting cases can be derived
Firstly , at high [CH:P] ,
8) f!l.ll = - {k2k)IlIH_l [OH') + L: II _ klm'l[B-] } [E) dt i.e. General base catalysis and reaction order zero w.r.t. formaldehyde. Carbanion formation ( reaction 4 becomes rate determining.
Secondly , at low [CH20 ]
9) f!l.ll = -k2kl(OH_I[OH-] [E) [CH2Ql
dt k- Il H20l[H20 )
i.e. Specific OH- ion catalysis . Reaction is first order w.r.t. formaldehyde. The condensation step ( reaction 5) becomes rate determining. 11
In both cases I reactions are first order w.r.t. [E]. At constant total formaldehyde concentration· rates of all reactions studied have been found to be directly proportional to the amount of NaOH added initially . The dependence of reaction rates on total amount of CH~O present is more complex.
Rate expression d{E} = -k l [E] [NaOHLdd«l where k! [NaOHLddod = k observed for 1st order disappearance of E. At low CH,concentrations the rates should be proportional to [CH~O]. At higher concentrations of CH~O the rate of anion formation ( reaction 4 ) becomes rate determining and reaction order in CH,O decreases to zero. This also applies to the process in operation at NCP.
2.2.1.3. Conclusions
(a) Rate determining step for the process is the rate of enolate anion formation (reaction 4).
(b) Rate is thus first order in acetaldehyde I first order in base and zero order in formaldehyde. It is general base catalysed.
(c) Rate decreases with increasing formaldehyde concentration due to acidity of formaldehyde hydrates reducing hydroxide ion concentration.
(d) Rates are directly proportional to NaOH concentration.
(e) Indications are that steric effects are not important and that rate follows ease of enolate formation.
(f) Rate of acetaldehyde self-condensation is < 3% of that of formaldehyde. 12 (g) The rate of reaction of enolate anion with formaldehyde is favoured because of the positivity of formaldehyde carbon.
(h) Even if self condensation of acetaldehyde occurs, it is known that acetaldol and even crotonaldehyde can undergo reversion in the presence of formaldehyde and base to give pentaerythritol. 13
2.2.2. CANNIZZARO REACTION
2.2.2.1. Mechanism
The Cannizzaro reaction occurs when no a-hydrogens are available, otherwise aldolisation occurs.
o
10) RCHO +OH- --+ R-C-H I' I OH
0 o ...
\~, II - I ~I II II I 11) R-C.-H + C-R - R-C-OH + H-C-R I "----"I I I OH H H
12) R-COOH + RCH,OH
(a) Fast reversible addition of OH-ion to aldehyde carbonyl gives intermediate X
(b) X then transfers a hydride ion to another carbonyl in the slow rate-determining step.
(c) The acid and alkoxide ion so formed exchange a proton to give the more stable acid anion and alcohol. 14
2.2.2.2. Kinetics
The reaction as above is third order; first order w.r.t. base and second order w. r . t. aldehyde. i . e. Rate ex [RCO] 2[OH-] with formaldehyde in high concentrations of base a fourth
order reaction can take place : ex [CH20]2[OH-]2
2.2.2.3. Conclusions
(a) Under normal conditions rate ex [a Ldehyde}? [OH-]
Reduced aldehyde concentrations will thus lower the rate proportionally to the square of the concentration, whereas for the aldol condensation it would lower the rate only proportional to the concentration, hence the need for excess formaldehyde to prevent side reactions.
(b) It is slow compared to normal aldolisations and does not compete significantly except at high temperatures.
(c) For formaldehyde it is· faster than the aldol-type condensation which leads to sugar formation, but slower than the crossed Cannizzaro in Penta formation. It would thus be advisable to start the reaction at a low temperature. 15
2.2.3. PENTAERYTHRITOL FORMATION
2.2.3.1. The specific reactions
For the normal reaction the ideal mechanism is as follows:
~ HOCH 2 ) CHCHO
17) HOCH2 ) 2 CHCHO via 13 and 15 - (HOCH2 ) 3CCHO
e 18) (HOCH2 ) 3 CCHO + CH2 (OH) OS - (HOCH2 ) 4C + HCOo i
CH20 + sOH
In reaction 13) to 17) the base is the catalyst, but in reaction 18) it is consumed stoichiometrically. This last reaction is irreversible and drives the reaction to completion. It does not proceed rapidly below about 40°C and pH 10.
2.2.3.2. Side reactions
The self Cannizzaro reaction of formaldehyde has already been discussed and is minimized by avoiding high pH, temperature and concentration.
The aldol type condensation is more serious from a product quality point of view because it results in sugar -like products (polyols) which eventually caramelize , imparting 16 The self-aldol condensation of acetaldehyde can give acetaldol, crotonaldehyde and potentially their aldol condensation products with formaldehyde. crotonaldehyde with base gives colour bodies. It is however known that these products can revert· in the presence of base and formaldehyde to give pentaerythritoI. The e f f ect can be minimised by ensuring adequate excess formaldehyde concentration and incremental addition of acetaldehyde so that its concentration does not buildup excessively.
Dipenta, poly-pentas and bispentamonoformal (BPMF) formation and mechanisms are also discussed in Laue's paper and in the Monograph by Berlow and Snow(1957) . Pentaerythritol Monoformal formation is also covered.
The presence of Methanol can lead to formation of various methyl ethers of pentaerythritol via mechanisms discussed in the above mentioned writings and is therefore to be avoided. 17
2.3. SUMMARISED CONCLUSIONS
The conclusions from all the foregoing have been summarised by Laue (1984) as follows :
CONDITIONS FAVOURS DISFAVOURS
MONO PENTA POLY PENTAS
FORMALS ACETALDEHYDE SIDE REACTIONS CR,O SIDE REACTIONS
EXCESS NaOH CH,O SIDE REACTIONS ALDOL CONDS. RATE CANNIZZARO RATE
EXCESS CHlCHO ACETALDEHYDE SIDE REACTIONS
LOW TEMPERATURE ALDOL OVER CANNIZZARO CH,O SIDE REACTIONS
HIGH TEMPERATURE CH20 SIDE REACTIONS CANNIZZARO
ACID WORK UP REVERSIBLE FORMAL ALDOL AND FORMATION AND CANNIZZARO REVERSION 18
2.4. RECENT RESEARCH SOME SPECIAL ASPECTS
Although the preceding information has been given as an overview of present knowledge the chemical mechanisms proposeed can by no means be regarded as final or immutable since new information is brought to light from time to time by current research, e.g. the pUblications by workers in East European countries. The model proposed by Belkin (1979) is one pertinant example and his results are shown graphically in the figure 2.1 below. Belkin's paper has been translated for the NCP team by Dr Yuri Tarakamov, ex Moscow State University, now living in South Africa.
Cz 6 a IZ M 0 10
.8 6 f A 0,_. ~~. Z I,. Z 0.15 0.55 0.95 A 0 3 5 7 9 1f M 10 40 70. T 2.5 3.75 5 MIS 2.5 3.75 5 MIB 0.15 0.55 0.95 A J 5 ,." B 1f M fO 40 70 T
Figure 2.1: Dependence of yield of pentaerythritol. C (a> and dipentaerythritol. C (s) on starting conditions at the central area. 1 - calculated values. 2 - experimental values
M - FormaldehydelAcetaldehyde mole ratio M/B - Formaldehyde/NaOH A- Acetaldehyde conc. ( molfl) T- Temperature deg. C
from Belkin (1979) 19
Also of particular interest is the paper by Trevoy and Myers(1963) dealing with the formation of Dipentaerythritol.By means of experimental work with radio isotopic acetaldehyde-Cl-l as a reagent in the Tollens condensation the authors have established that monopentaerythritol is an intermediate in the formation of dipentaerythritol. 20
CHAPTER 3 CHEMICAL REACTORS FOR PENTA PRODUCTION
This chapter first gives a brief introduction to some of the principles of chemical reactor design in general and then follows with a discussion of the various types of equipment used in the particular case of Pentaerythritol preparation. Finally, brief descriptions of some industrial processes are given. The well known books by Levenspiel (1962,1979) serve as suitable references. Recent feature articles by Dickey (1991) and Fasano and Penney (1991) give an introduction to stirred tank reactor concepts.
3.1 CHEMICAL REACTORS AN INTRODUCTION
Equipment in which homogeneous chemical reactions are carried out can be one of three general types; batch, steady-state flow, and unsteady-state flow or semi-batch reactor. The illustrative sketches in figure 3.1. are adapted from Levenspiel.
3.1.1 THE IDEAL BATCH REACTOR
If one were to write a material balance for any component species A in this case both input and output would be = 0 and only disappearance and accumulation would have values. The balance equation then becomes :
+(rate of loss of A) = -(rate of accumulation of A)
and the general performance equation is
x, dX t=N J n .'1 ADo (-r ) V A 21
which for constant fluid density may be simplified to
XA ax, t=C --" :\00 -r J .?j
Equipment used as a reactor when reaction takes place in liquid form is commonly some kind of stirred tank reactor (STR) and thus for a batch process it would be termed a Batch stirred tank reactor (BSTR).
Most real BSTRs operate more often as variable volume batch or semi-batch reactors (SBSTRs), where at least one reactant is gradually added over a period of time as the reaction takes place, or different reactants or conditions are involved during the complete processing cycle. In general the characteristic STRs including SBSTRs (and also the CFSTRs to be discussed next) make use of stirring and mixing effects provided by fluid mixing and sparging systems.
3.1..2 STEADY-STATE MIXED FLOW REACTOR
The material balance equation for any component in a steady-state flow reactor e.g. a Constant Flow Stirred Tank Reactor, accumulation being = 0, becomes:
input = output + dissapearance by reaction
The performance equation for a mixed flow reactor is
-----F X.'-. F AO -rA 22
For constant density this can be written
-='t=~v VC V F.~o
For liquid processes the Constant Flow stirred Tank Reactor (CFSTR) is commonly used. In the design and operation of real CFSTRs extensive use is made of agitation mixing and sparging systems.
3.1.3 STEADY-STATE PLUG FLOW REACTOR
In a plug flow reactor (PFR) I fluid composition varies from point to point along a flow path. The overall material balance is the same as for mixed flow but the performance equation becomes
~= IX;. dX,; FAO 0 - I.; 23
Composition at any point is unchanged with time Feed-~~~f '~ - ?C::~ ~i i~---Product
(b)
B'\ ~ B(
Volume8 and /' composition change (c) Volume changes)U but composition is unchanged (d) Volume is constant / but composition ./ changes (e)
Figure 3.1 : Classification of reactor types: (a) The batch reactor. (b) The steady-state flow reactor (c).(d) and (e) Various forms of semi-batch reactor from Levenspiel (1962) 24
3.1.4 IDEAL AND NON-IDEAL FLOW IN REACTORS
The models described above represent ideal situations. However real equipment usually deviates from ideals. In "Chemical Reaction Engineeringll and also in his "Omnibook" Levenspiel mentions the overall contribution to reactor - contacting or flow patterns of three somewhat interrelated factors. These are the residence time distribution (RTD) of material flowing through the vessel,the state of aggregation of the material, its tendency to clump and for groups of molecules to move about together, and the earliness or lateness of mixing of material in the vessel. Concerning the second of these factors it can be shown that fluids can be described as either macrofluids in the case of extreme aggregation or as microfluids when little aggregation occurs.
Significantly it can be said that for separate reactant streams entering a reactor where a fast reaction is concerned these three physical factors will control reactor behaviour and chemical kinetics becomes unimportant. 25
3.2 REACTORS FOR THE PREPARATION OF PENTAERYTHRITOL
The chemical reaction process used for Pentaerythritol preparation takes place in a single liquid phase and the product is also quite soluble in the same reaction mixture at the conditions under which that mixture is maintained. The effective volume change at constant atmospheric pressure can thus be regarded as negligible. Laboratory scale preparation is carried out as a batch process and industrial processes are also commonly so designed. This can thus be ideally considered as a Constant Volume Batch Reactor. Semi-batch type processes as well as continuous mixed flow arrangements have been used.
The book by Berlow et. al.(1957) gives brief descriptions of reaction methods used for the preparation of pentaerythritols. There are various procedures in which acetaldehyde and alkali are added toa batch of formaldehyde solution with some form of stirring and temperature control over periods of time ranging from less than an hour to 5 or 6 hours. These processes can be regarded as taking place in semi-batch or variable volume batch, i.e. unsteady-state flow reactors. Processes in which the reagents are continuously mixed and passed through a reaction zone or vessel, i.e. steady-state flow reactors are also mentioned. Various reagent concentrations and molar ratios have been reported. 26
3.3 SOME INDUSTRIAL PROCESSES
The following brief descriptions of processes in commercial operation have been adapted from the NCP literature survey
by stoltz and Salemi (1979a )•
3.3.1 HERCULES POWDER CO. PROCESS
The reactor is a stirred tank with internal cooling coils resting on load cells. Formaldehyde solution and lime as a 25% slurry in water are charged to the reactor. Acetaldehyde is then charged incrementally from a weigh tank by nitrogen pressure. The temperature is critically and carefully controlled so as not to exceed SooC at any time. The reaction process takes 2 hours. The complete reaction mixture is then pumped to stainless steel precipitator tanks with agitators and coils. Sodium carbonate is added to precipitate calcium carbonate, forming sodium. formate. The mixture of sodium formate, pentaerythritols and suspended calcium carbonate is pumped to a continuous rotary filter where carbonate and insoluble polypentaerythritols are removed. Further product recovery and purification is essentially outside of the scope of the present work but the flowsheet , figure 3.2. shows the main operations. No information on formaldehyde to acetaldehyde ratios used in this plant could be obtained but in their report stoltz and Salemi assumed a value of less than 6:1. 27
_~ ..-aut~ --II-.n~==:::===f.r;===~-:::::--:=::~":::::---'[::l:btl~:r----;=:=
In.. ..- ..
Ie-
Figure 3.2 Flowaheet on the plant proceaa of Hercules Powder Company. 28
3.3.2 JOSEF MEISSNER PROCESS
Josef Meissner is a process engineering and contracting company who have supplied a number of chemical production plants to clients, including at least five pentaerythritol plants. The flow sheet is shown in figure 3.3. and a short description of the reaction processes follows.
3.3.2.1 Meissner process with Sodium Hydroxide catalyst
This is a continuous reaction process, using caustic soda as catalyst. The formaldehyde solution, acetaldehyde at 98 to 99 % strength, and 25 % caustic are fed into the primary reactor under intense mixing and cooling when the initial aldolisation supposedly takes place. After a certain time (condensation 1 the mixture passes through a calming zone (condensation 2 in which the reaction is completed, i.e. the Cannizzaro.
A high formaldehyde to acetaldehyde mole ratio is used, resulting, according to the authors in a high formals concentration, hence the need for a formals splitting step. Neutralization with formic acid is carried out in a hold up tank, after which further recovery operations take place.
3.3.2.2 Meissner process with Calcium Hydroxide catalyst
The process differs from the sodium hydroxide option in the following respects.
(l) The calcium hydroxide is added to the reactor as a suspension.
(2) The solubility of calcium formate differs from that of sodium formate and some calcium formate crystallizes 29 in the heated solution during evaporation. For this reason the recovery process is somewhat different, the calcium formate being separated first by crystallization, before the pentaerythritol is crystallized.
Further treatment of residual mother liquor is similar to that in the sodium hydroxide process. The calcium formate is treated at high temperature to form calcium hydroxide which can then be recycled as catalyst to the reactor. 30
I 1
J 'i i (I) II) ii ..,o (I) -s:
I 1
= I I ,I: "n 31
3.3.3 PERSTORP PROCESS
Information on this process plant operated by the Swedish firm Perstorp is mainly derived from data obtained by c. Manushewitz (1972) of NCP on a visit to the plant. An important feature of this process is the two stage reaction arrangement.
The initial reaction takes place in a cold temperature reactor ( 1st stage ). The formaldehyde charge is cooled to O°C and sodium hydroxide catalyst is added over a two minute period during which the temperature rises to BOC. Acetaldehyde reagent is then added at a controlled flow rate over 1 hour such that a final temperature of 31°C is attained. The batch i~ then pumped to an after-reactor ( 2nd stage ) in which the temperature is further increased to 40°C. After completion of reaction and formic acid neutralization the excess formaldehyde is recovered in a distillation column and product recovery takes place in evaporation, crystallization and solids separation operations.
Although data is rather incomplete the drawings supplied by Manushewitz seem to show that the Perstorp primary reactor has an agitation and reagent sparging system of refined design. The original NCP pentaerythritol process was based on Perstorp technology. This plant was later replaced by one using technology from International Minerals Corporation IMC). The IMC type plant is presently still in use at the Germiston site and the NCP reactor system will be described in greater detail in the following chapter. 32
CHAPTER 4 THE NATIONAL CHEMICAL PRODUCTS' PROCESS
The following is a brief description of the reaction process currently used for Penta production at the Germiston works of N.C.P. The equipment and reactor design considerations and the chemical process procedures will be discussed, with special attention to the disadvantages inherent in the reagent addition arrangements. T~e process is based on technology obtained from licencors International Mineral and Chemical Corporation (IMC).
4.1 THE CHEMICAL REACTOR
The NCP plant consists of a primary stirred tank reactor, an SBSTR, with an external temperature control circuit, and two secondary stirred tanks which are used alternately.This makes it possible to prepare a batch in the primary reactor while the previous batch reaction is still nearing completion in one of the secondary vessels.The process goes to approximately 90 % completion in the primary reactor and this equipment will now be considered in greater detail.
4.1.1 NCP PRIMARY REACTOR AND COOLING CIRCUIT
The primary SBSTR system consists of a stirred tank with an external recirculating stream passing through a plate heat exchanger for reaction temperature control.The general arrangement and principal dimensions are shown in figures 4.1 and 4.2. From the design data and dimensions and using correlations given by Bowen (1986), who has published various papers on fluids mixing, estimations of agitation parameter values have been obtained for the stirred tank. These parameters are Reynolds number of mixing, Power number, Pumping number and hence power drawn, torque, "agitation intensity", fluid velocity and blend time. For the pumped cooling circuit an estimation of pressure 33 drop under operating conditions was obtained by direct measurement. A system curve for the circuit was calculated, using dimensions shown in the figures and reasonable assumptions for equivalent lengths of pipe fittings, valves and "relative roughness" of interior surfaces.
By interpolation from the pressure reading on the curve the flow rate was estimated and thus also the pipeflow Reynolds number and recirculation turnover time for the reactor.The procedures suggested by Bowen and other relevant calculations are shown in the Appendix 1 . The resultant values are shown as follows.
FLUID FLOW PARAMETERS
STIRRED REACTOR PUMPED CIRCUIT
Re(ml 5.09 * 105 Re 1.09 * lOs Np 6.2 q 0.·05 m3 S·l Nq 0.96 u 1.59 m s' 3 P/V 0.98 kW m- eTO 508 s T/V 0.14 kN m-2 sol u b 0.224 m eB 19.79 s
N·I 7.35
It must be emphasised that Reynolds numbers or similar parameters for pipeflow and mixing cannot be compared directly but on the respective scales the numbers obtained both indicate a degree of turbulence well beyond the transition range.
It is also not valid to make direct numerical comparisons of the torque/volume vs. discharge flow relationship for a mixing system with a pumping system or pump performance curve but in this case it can be said that the agitator 34 discharge flow is principally in the radial direction and th~ flow rate is typically 2/3 of the valu~ for an axial flow or pitched blade turbine with dimensionless power number of similar magnitude, i.e. whereas the circulation pumping system is des i.qned : as the name suggests, the agitation system has more of a shearing as opposed to a flow inducing character. Both of these elements would be indicated for the purpose of reagent dispersion, the latter on the macro, the former on the micro to molecular scale. 35
Figure41'NCP'.. pnmary reactor- syst " and reagent charging ( 0" em. ~oohng circuit . tmenstcna in mm ) 36
4.1..2 REAGENT ADDITION AND SPARGING SYSTEM
The addition of acetaldehyde to the reactor is effected via a dip pipe with fifty 6mm diameter holes.
The acetaldehyde is weigh batched in a charging vessel under 250 kPag pressure and from there passed at a fixed rate through a control valve into the reactor which is at atmospheric pressure. At 250 kPa pressure acetaldehyde boils at 70°C i.e. well above ambient temperature, but for 101 kPa (atmospheric pressure) the boiling point is 20.2°C. This means that depending on such factors as ambient temperature and heat transfer to material in the dip pipe the acetaldehyde reagent could enter the reactor either as a liquid or vapour, or as a mixture of both. The last case is very probably the situation most of the time. Measured values for the temperature of the dip tube at the point of entry into the reactor have been recorded varying between 11°C and 20°C.
Acetaldehyde enters the reactor at a controlled linear rate of 950 kg over 80 minutes 'which is equivalent to 4,89 molest sec. or 0.253 lis as liquid. Velocity in a 50 mm I.D. tube would then be approximately 0.12 mls with a Reynolds number of about 17000. If the liquid passing into the reactor were equally distributed among the 50 holes then at each hole the orifice Reynolds number would be about 3000. If however the 4.49 moles/sec. of acetaldehyde were in the gaseous state at 20°C or >20 oC, the volume would be much larger and also the velocities and corresponding gas orifice Reynolds numbers would be of the order of 9000.
For gas sparging into liquid e.g. for aeration of fermentation reactors, it is generally accepted that ideally the orifice Reynolds number value should not be below 6000 ( Perry, 1985 ). Thus it would seem that were 37 the flow to be evenly distributed between the 50 holes in the dip tube, for a liquid the Reynolds number would be extremely low whereas if the acetaldehyde were gaseous then the magnitude of the same parameter should be sufficient.
It would however seem fairly obvious that the above condition, i.e. even distribution between the holes, should not be expected to apply. This should be ruled out simply by virtue of the difference in static head between the holes. It can then be inferred that the acetaldehyde addition arrangement design is severely lacking in any scientific basis. The reagent dispersion effect can then also not be regarded as optimal. 38 N
REAGENT DIP PIPES G. 8 I Ce I Osh \ · I I·
z
I · I ,j Q. • ~ '0 · U :z:: I · I
l 0" ~
I. Odp .J 0,
Figure 4.2 : NCP Primary Pentaerythritol Reactor. ( dimensions given overleaf) 39 Penta primary reactor principal dimensions
B Baffle width 0.25 [m]
C Agitator turbine off-bottom distance 1.37 [m]
Cn Baffle distance 0.1 [m]
Cs Interstage distance 1.5. [m]
DA Agitator impeller diameter 1.2 [m]
D~ Dip pipe distance (radial) 2.0 [m]
D~ Shaft diameter 0.18 [m]
DT Vessel diameter 3.2 [m]
H~ Dip tube height (of holes) 2.3 [m]
HL Liquid height 3.62 [m]
L Blade length 0.3 [m]
N Shaft speed (65/60) = 1.083
w Agitator Turbine blade width 0.25 [m]
z Number of agitation stages 2 [Number] 40
4.2 REACTION PROCESS CONDITIONS
4.2.1 MOLE RATIO
The ratio used is formaldehyde acetaldehyde caustic
:: 6. 7 1 1.15
The formaldehyde and caustic are thus in excess and
acetaldehyde is the limiting reagent.
4.2.2 REAGENT STRENGTHS
The chosen reagent and caustic catalyst strengths are
formaldehyde 16.7%
acetaldehyde the pure reagent ( 99% min.) is used
catalyst 43% caustic soda solution
REACTION TEMPERATURE
The operating temperature is controlled at 40°C 41
4.2.4. REAGENT ADDITION PROCEDURE
The Formaldehyde solution batch is charged to the primary
stirred reactor which also has an external cooling circuit for temperature control. An initial 10% of the equivalent
amount of caustic catalyst solution is then charged into
the reactor, whereupon addition of the Acetaldehyde reagent
is commenced. The Acetaldehyde addition rate, as well as
that of the caustic, is so controlled that complete
addition takes place over a fixed period of 80 minutes,during which the reactor is continuously stirred
and the temperature .controlled by means of the cooling
circuit.
After completion of reagent addition, at which time the
reaction will also have progressed to near completion, the
batch is transferred to a secondary stirred tank where
reaction progresses to completion. The excess caustic is
then neutralised by addition of Formic acid and the batch
transferred to the product recovery systems. 42
CHAPTER 5 PROBLEM, STRATEGY AND EXPERIMENTAL DESIGN
Following on the discussion of the NCP reactor in chapter 4, the first part of the present chapter consists of a
summing-up and statement of the primitive problem. The
second section gives a statement of the primary objective of the experimental work while drawing attention to some
practical difficulties to be overcome. The third part of
the chapter sets out in broad terms the way in which the
objective was to be achieved, the experimental design.
5.1 THE NCP REACTOR PROBLEM
From the information presented in chapter 4 it is fairly
obvious that in the case of the NCP plant mixing and the
degree of turbulence in the reactor and recirculation
system should be sufficient for pentaerythritol formation.
The reagent addition and sparging systems are however not
optimally designed, which fact constitutes the primitive
problem, and for the purpose of reactor optimisation this
aspec~ is the one on which attention will be focused in the
following pages. It is certainly possible to propose an
alternative or improved sparging system using a practical
approach. It is however important that any major technical
changes to industrial processes or equipment should be
SUfficiently justified by practically demonstrable
information or concrete results. 43
In the case of the NCP Penta reactor the original intention
was to carry out some form of full scale plant
experimentation, including testwork involving the use of
modified equipment. The cost and associated risks of such
modification and plant trials could however not be
justified, the position being indeed that a "Catch 22"
situation existed. There was insufficient "hard" data to
justify the plant alteration and for the same reason the
cost of plant scale testwork could also not be considered.
There was also the added disadvantage that conditions of
plant instability could make it quite difficult to achieve
significant results f.rom such experimentation.
5.2 OBJECTIVE STATEMENT AND PRACTICAL CONSTRAINTS
Because of the foregoing considerations it was decided that
the problem would be addressed by attempting to perform
some type of laboratory or pilot scale experiments, the
results of which would hopefully, as a primary objective,
indicate the effect of improved reagent sparging. To
successfully achieve this the use of suitable "Chemical
Engineering laboratory" or pilot-scale apparatus would have
been highly desirable. Unfortunately however such equipment
was not readily available and would have been prohibitively
expensive to manufacture or otherwise procure. The only
equipment available was the bench-scale rig used previously
by Stoltz and Salemi (19791,) and more recently by Simon
). (19913 This was initially not thought suitable for the 44 specific purpose but after some preliminary testwork it was possible to modify the apparatus so that significant results could be obtained. The technical details will be described later. The experimental strategy was devised to allow for the technical constraints as shown in the following paragraphs.
5.3 EXPERIMENTAL DESIGN
The experiments were planned so as to show the effect of variation in fluid flow and sparging characteristics at controlled reagent mole ratios on the formation of Pentaerythritol and its by-products. The temperature of the reactor was to be controlled at a constant fixed value for all the tests, i.e. it was not to be considered as a variable parameter. The independent variables under consideration were thus the mole ratios and the fluid flow or reagent dispersion character.
It was decided to carry out the tests with the Formaldehyde/ Acetaldehyde mole ratio variable at three different levels. The values chosen were 9.0, 6.5 and 5.3. The intermediate value, 6.5 is approximately the mole ratio commonly used in the NCP industrial scale process. The extreme values are arbitrarily chosen but practically similar to those suggested by Van Niel (1991) in reply to
proposals by the author (Jennery, 1991<). 45 The fluid flow or dispersion parameter was varied by manipulating the combination of agitation intensity and reagent sparging mode as far as it was possible within the constraints imposed by the· practical limitations of the equipment available at the R&D laboratory. These difficulties had been referred to in a previous report (Jennery, 1991<). Because of the aforesaid constraints it was considered prudent to aim at achieving two levels only for this parameter, these being the extreme values practically possible. They could be designated simply as "well mixed" and poorly mixed or "not well mixed", this being preferable under the circumstances to any futile attempts at quantitatively estimating the fluid flow parameters.
The dependant variables to be evaluated were quantities descriptive of the type and degree of chemical reaction and product formation taking place in the experiments. These values were to be assessed by sampling the liquors on completion of the chemical reaction process and chemical analysis of the samples. The time allowed for reaction was to be'the same for all the experiments and the process was to be stopped by neutralization of the caustic catalyst in the samples, thus "freezing" the species composition profiles for analysis.
The data obtained from the experiments was to be processed and presented in a suitable manner to show the effects of 46 manipulation of the independent variables on the dependant chemical product species distributions.
Thus to conclude, it has been shown in this chapter what the primitive problem was, how it was to be solved, i.e. the immediate primary objective of the experimental work,
and the manner in which it was intended to show the effect
of changes in the values of independent variables on dependent variables, i.e. the experimental design. 47 CHAPTER 6 THE EXPERIMENT
This chapter consists of a description of the experimental
details of the testwork which was undertaken to achieve the
objective according to the design given in chapter 5.
The following sections are thus descriptive of the test
materials, the experimental apparatus and the experimental
procedure. The chemical analyses and the examination of
test results are briefly introduced but not described in
detail as they fall outside the scope of the present work
and this chapter respectively.
6.1 TEST MATERIALS
The reagents used for the test work were the following
FORMALDEHYDE Samples of this reagent were taken from
the NCP Formaldehyde production plant. The material,
usually at a concentration of ca. 40% was diluted with
water to a strength of 15% mjv.
ACETALDEHYDE Acetaldehyde from the NCP plant was kept
at low temperature (ca. -70 0C) in the laboratory. The
reagent was thus used as a liquid.
CAUSTIC SODA SOLUTION Plant reagent from Penta
Production Department was taken and diluted with water to
give a solution with a concentration of 36% mjv. 48
6.2 EXPERIMENTAL APPARATUS
The apparatus used for conducting the testwork was the
laboratory glassware test rig used by Salemi and Stoltz for
their investigations at NCP during 1979 and also recently
by Simon (1991,) in his work on reaction temperatures. This
test rig was further developed and improved by the author,
with the technical assistance of Miss TC Mvelase of NCP
R&D, as described below. A more detailed report on this
aspect was prepared separately (Mvelase,1991)
The original arrangement of Stoltz and Salemi employed an
adjustable reciprocating pump for reagent delivery. It was
at first attempted to use similar equipment f~r the present
work. It soon became apparent however that to achieve the
accurately controlled acetaldehyde delivery rates required
for the testwork the equipment and technique was
inadequate. This was due to the alternate compression and
expansion of the liquid in the pump cylinder causing it to
vaporise in spite of all attempts to cool the material and
apparatus down sufficiently to counteract this effect.
Finally, after many preliminary trials, effective delivery
of the acetaldehyde was achieved by means of a glass
syringe through 1mrn diameter tUbing. The caustic soda was
dispensed via the reciprocating metering pump as described
by Salemi and stoltz and also used by Simon and shown in
figure 6.1. The acetaldehyde syringe pump delivery rate 49 and volume of acetaldehyde was fixed for all the experiments. All other quantities, e.g. rate of caustic addition, were adjusted accordingly. 50
STIRRER MOTOR,
METERING THERMOCOUPLE THERMOMETER PUMP -. I <, HEATER. -. ... PUMP I ~~ .. - ~~ -0 1 - - I ~ i- - f~ U i- 60 . - t - ..... - ~ ~ - ~~~ f-. /V /' - ~ - f- /~ -NaOH / V/ /,/ ~/~-/ YiATERBATH -- V~/ / ./ ~ ... /--. V ACETALDEHYDE" REACTOR
.figure 6.1 : Laboratory bench scale reactor for pentaerythrltol preparation a8 used by Stoltz & Salem1(1979) and by Simon (1991) 51
,"y~"...... ;..;.." /6y·fi ....,. - "'----... ,-.. "U""P ~ STIRRER MOTOR I
METERING _ ....._ .. THERMOCOUPLE THERMOMETER PUMP I -
WATER BATH "
Figure 6.2 : Bench scale test rig tor reactor parameter experiments with controlled stirring, reagent sparging and dispersion using a syringe pump tor acetaldehyde teed.
120 mm
, • 20 mm i 115 mm ___....L... -/J
{ i 45 mm dia. I 4 blade impeller II 50 mm
~____<.'.>----..40----
Figure 6.3: Bench scale stirred tank reactor showing principal dimensions. 52
6.3 EXPERIMENTAL PROCEDURE
For each test the reaction ~pparatus was set up and the required volume of 15% Formaldehyde introduced. The
agitator was started with the motor setting adjusted to
give the required shaft speed.
The temperature control system pump motors, thermostat and
heater were started and adjusted to give the required constant reactor temperature. When this was reached the caustic addition could commence. The metering pump,
previously adjusted for controlled rate addition, was
started before the syringe pumps, so that the reactor could
be preconditioned with 10% of the caustic. The syringe
with Acetaldehyde could then be started. Reagent addition was continued at constant rate until completion. Agitation
and temperature control systems were allowed to continue
operation up to 6 hours from commencement of Acetaldehyde
addition. A representative sample was then withdrawn and
the motors stopped. The sample was neutralized with Formic
Acid to ca. pH5 and submitted to chemical analysis. If,
for practical reasons analysis was not immediately possible then the sample would be stored in the freezer compartment
of a refrigerator. 53
6.3.1 ADJUSTMENT OF INDEPENDENT VARIABLES
6.3.1.1 Agitation and reagent dispersion
The reactor agitation and/or dispersion conditions were adjusted at two levels in the following way :
Well mixed reactor (WMR) The motor was adjusted to give an
agitator shaft speed of 1000 r.p.m. (16,67 S·I) which
resulted in a degree of turbulence, (Reynolds number of mixing of the order of some thousands) and no vortexing observed. The acetaldehyde delivery tube was dipped 25mm
below the surface of the reaction liquor.
Not well mixed (NWR) Motor adjustment to give shaft speed
50 r.p.m. (0,83~l), (Reynolds number of perhaps 100 - 500).
The acetaldehyde tube was suspended 25mm above the surface,
i.e. the reagent entered the reactor as droplets falling
into the liquid surface.
6.3.1.2 Mole ratio (M.R.)
The formaldehyde solution quantity was adjusted to give
three different reagent mole ratios. These were
Formaldehyde Acetaldehyde: Sodium Hydroxide
9.0 : 1 : 1, 6.5: 1 : 1 and 5.3 : 1 : 1.
The quantities of the two last named and, of course, the
delivery rates, remained the same for all the tests. 54
6.3.2 REAGENT ADDITION RATE
The rate of addition of acetaldehyde by the syringe method
was dictated by the capacity of the syringe and by the stroke and period. The model used, and there were two, delivered 20mf of the reagent at specific gravity of 0,85
(-70 0C) over a period of 33 minutes. Two syringe fulls were thus delivered over a period of 66 minutes. This
quantity was thus 0,80 moles. The rate of caustic soda
addition was adjusted to be equivalent, with the provision
that the addition of this reagent was given a 10% (6,6
minutes) head start.
6.4 CHEMICAL ANALYSES
The samples were analyzed by the methods outlined in an
internal NCP report (Simon, 1991.} and currently being
published (Simon et. al., 1993). The species determined
were mono-pentaerythritol, di-penta, bis-penta mono-formal
(BPMF) and a miscellany of minor constituents collectively termed "unknowns".
6.5 CRITICAL EXAMINATION OF RESULTS
The chemical analyses were used to compute the relative
amount of each of the species as a percentage of the total
of products for each sample. The yields of mono penta, di- 55 penta and BPMF, based on Acetaldehyde as a limiting reagent, were also computed, as were the selectivities for
the species. The mean results for the sets of tests were
computed. The results are presented in the form of Bar
Graphs which are given in the following chapter. Calculation formulae and tables of results are given in the appendix 2.
6.6 TESTWORK SETS
The numbers of the designed tests performed and whose
results are presented in the following chapter are as follows
WMR, MR = 5.3 1 1 4 Tests NWM, MR = 5.3 1 1 4 Tests WMR, MR = 6.5 1 1 4 Tests NWM, MR = 6.5 1 1 5 Tests
WMR, MR = 9.0 1 1 3 Tests NWM, MR = 9.0 1 1 2 Tests
The designations "WMR" and "NWM" above denote "Well Mixed
Reactor" and "Not Well Mixed" respectively, "MR" being
"Molar Ratio". 56
CHAPTER 7 RESULTS AND DISCUSSION
In the first section of this chapter the results of the
experimental work designed and described in the previous
two chapters are presented graphically. Discussion of these results follows and finally conclusions and inferences are
given.
7.1 RESULTS
The Chemical Analyses were used to calculate species
distributions,. yields and selectivities. Calculation formulae and table of results are given in Appendix 2. Mean results are presented in graph form, in figures 7.1 to
7.10 below. Explanatory notes are given in the following
paragraphs.
Figure 7.1 shows the effect of dispersion mode and mole
ratio variation on the production of mono-penta.
Figure 7.2 shows the effect of variation of the same
parameters on by-products formation. The quantities shown
are the masses of the various species expressed as
percentages of the total mass of products. Mean totals of
mono- and di-penta are given in figure 7.3. 57 Figures 7.4 and 7.5 portray the amounts of the various product species expressed as molar yields based on acetaldehyde, also showing the effect of the experimental parameter variations.
Figures 7.6 and 7.7 show the effect on the molar selectivities.
Figures 7.8, 7.9 and 7.10 show the range of the results of which the means were used to produce the graphs given in figures 7.1 to 7.7. 58
MR 5.3 WELL MIXED
MR 5.3 NOT MIXED
MR 6.5 WELL MIXED
MR 6.5 NOT MIXED
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED
o 20 40 60 80 100 % m/m _ MONO-PENTA
Figure 7.1 : Monopenta as % m/m of total products.
&'§\§\\\W§$Sfu§\§$$§\\\S\SW$\$\\\\\SS\S\S\$§%S§S§$SS§%S\\S$\SS$\S\SSS\§S\\\\$$$§$§\\\\i MR 5.3 WELL MIXED
MR 5.3 NOT MIXED MR 6.5 WELL MIXED - MR 6.5 NOT MIXED
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED
o 2 4 6 8 10 12 % m/m
_ Ol-PENTA i I B.P.M.F. _ ·UNKNOWNS.
Figure 7.2 : Penta by-products as .. m/m of total products. 59
I MR 5.3 WELL MIXED b I I MR 5.3 NOT MIXED E I MR 6.5 WELL MIXED
MR 6.5 NOT MIXED ~
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED t I 0 20 40 60 80 100 % m/m
---..: MONO- • DI-PENTA
Figure 7.3 : Pentaerythritols (Mono • 01) as % m/m of products.
MR 5.3 WELL MIXED
MR 5.3 NOT MIXED
MR 6.5 WELL MIXED
MR 6.5 NOT MIXED· • _
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED
o 0.2 0.4 0.6 0.8 1 Yield
- MONO-PENTA
Figure 7.4 : Yields based on Acetaldehyde Mono-Penta : Moles per mole 60
MR 5.3 WELL MIXED .\\\\\%\\\\\\\\\\%\\\\\\\\\\\\\\\\\\\\\\\\\\\\\1 I I MR 5.3 NOT MIXED r\\\\\\\'\\\\\\\)\\\?""
MR 6.5 WELL MIXED
MR 6.5 NOT MIXED
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED
o 0.02 0.04 0.06 0.08 0.1 Yield
_DI-PENTA ~B.P.M.F.
Figure 7.5 : Yields based on Acetaldehyde Di-Penta and B.P.M.F. : Moles per mole
MR 5.3 WELL MIXED
MR 5.3 NOT MIXED
MR 6.5 WELL MIXED
MR 6.5 NOT MIXED
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED
o 10 20 30 40 50
_ MONO-PENTA
Figure 7.6 : Molar Selectivity for Mono-Penta 61
I MR 5.3 WELL MIXED ~\\\\~\\~\\~~\\\\-\-~~\\~\\-\%\\\\\~~'§ I ! I MR 5.3 NOT MIXED _\\\\\\%\\\\\\\\\\\%\\\\\\'il
\ MR 6.5 WELL MIXED
MR 6.5 NOT MIXED
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED
o 0.02 0.04 0.06 0.08 0.1
_DI-PENTA
Figure 7.7 : Molar Selectivity for Oi-penta .
MR 6.3 WELL MIXED ~,\, .\\ .\\\\\ .\\\\\\\'01
MR 6.3 NOT MIXED .\\' .\\\\\' .\\'11
:
~ MR 6.6 WELL MIXED
MR 6.6 NOT MIXED
MR 8.0 WELL MIXED
MR 8.0 NOT MIXED .\ .\\\\\ .\\\' I I o 20 40 60 80 100 " mlm - MONo-PENTA LOW ~ HIOH
Figure 7.8. Range of re.ults Mono-penta •• " mlm of total products. 62
MR 6.3 WELL MIXED
MR 6.3 NOT MIXED
MR 6.6 WELL MIXED
MR 6.6 NOT MIXED
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED
o 10 15 20 ,. m/m
_ DI-PENTA LOW ~ DI-PENTA HIGH
Figure 7.9 : Range of resulta : DI-Penta aa ,. m/m of total producta. .
MR 6.3 WELL MIXED
MR 6.3 NOT MIXED
MR 6.6 WELL MIXED
MR 6.6 NOT MIXED
MR 9.0 WELL MIXED
MR 9.0 NOT MIXED
o 2 4 6 8 10 ,.m/m
CJ BP.M.F. LDW ~ B.P.M.F. HIGH o -UNKN.· LDW o ·UNKN: HIGH
Figure 7.10 : Range of re.ult. : BP.Mr. and ·Unknown.- a. ,. mlm of total products. 63
7.2 DISCUSSION
For interpretation of these results the following caveats should be borne in mind :
The experiments were carried out in laboratory glassware.
The so-called "Test rig" is NOT a scaled-down NCP Penta
reactor, neither is it a pilot Plant or simulation in chemical engineering terms. This difficulty was indicated in the introduction and also previously in internal NCP
communications by the author (Jennery,1991J and L.A. Van
Niel(1991) .
In spite of the. above mentioned problems it would seem that
the results clearly indicate a pattern of product species
distribution in the reacted liquors which can only be ascribed to the differences in mixing and dispersion modes.
The effects are more pronounced for the lower mole ratio situation and can be summarized as follows :
WELL MIXED REACTOR VS NOT WELL MIXED
MONO PENTA Little effect on Mono Penta formation.
DI-PENTA Increased Di-Penta formation. 64 MONO PLUS DI-PENTA Overall increased production
ofthe two products together.
Less BPMF formation.
"UNKNOWNS" Diminished formation of these compounds.
COLOUR Pronounced diminished formation of coloured species.
7.3 CONCLUSIONS
Interpretation in terms of chemical reaction mechanisms is
not intended here, but the results seem consistent with
theory. That the increased Di-Penta production probably
follows from the enhanced mono formation seems to be a
reasonable speculation, the latter being a precursor of the
former. This has also been indicated in the work by Trevoy
and Myers (1963) and in that of Belkin (1979) referred to
in chapter 2. In general it seems that the engineering of
a situation in which each Acetaldehyde molecule could "see"
more of other species sooner has had an effect, as
predicted. In experiments at this minute scale it would be
virtually impossible to observe any influence due to other
transport phenomena such as species diffusivities. It is 65 to be expected that similar effects to those in the
laboratory experiments will take place, and be amplified,
in the case of manipulations on the industrial scale.
Conclusions of this nature, although perhaps obvious, have
not been mentioned before, at NCP or in literature surveyed
to date (Simon,1991h ). Certainly one could safely assume that experimental results of this type have not been
published before. These findings should thus be regarded as a significant contribution to the general universal body of
knowledge concerning the Pentaerythritol condensation
process. 66
CHAPTER 8 : REAGENT SPARGING SYSTEMS
In the first section of this chapter .» reagent sparging systems for pentaerythritol production reactors are discussed in the light of the experimental results and conclusions of chapter 7. Secondly design recommendations for reagent sparging equipment in the case of the NCP reactor specifically are given. section 8.3 gives a brief discussion of alternative methods of reagent sparging. In conclusion a specific recommendation is given for modification of the NCP plant.
8.1 PENTA REACTOR SPARGING SYSTEMS PRACTICAL CONSIDERATIONS
The experimental results as presented graphically in the preceding chapter show that improved reactor mixing and sparging leads to greater yields of pentaerythritols and diminished yields of side products for equal formaldehyde/acetaldehyde mole ratios. It is thus not unreasonable to suppose that improved mixing in a commercial reactor would make it possible to yield suitable penta products at lower mole ratios. This would increase the capacity of the reactor. Or, if the mole ratio were kept at a higher level the effect would be to yield less side products. Either way there would be a reactor optimisation effect. Another advantage would be the
virtual cessation of adverse colour development in the 67 reaction liquor. If the, mole ratio in the present NCP reactor could be lowered from the current ca. 6.5 to a level of say 5.8 the production capacity would rise by about 12% or roughly 1.9 tons of penta per day. Assuming recovery at a rate of 80% this is equivalent to 1.5 tons and at a profit margin of R2 500 per ton it would be worth in the region of R1.4 M per annum.
The critical examination in chapter 4 of the NCP reactor design shows that although the liquid agitation is quite vigorous the reagent feeding arrangement seems to be very inadequate. The sketch, figure 8.1 shows the dip tubes arrangement and the principal dimensions are given in the table. The following section gives process design details for the construction of optimised reagent sparging systems for the NCP reactor. This includes proposals for both the acetaldehyde and the caustic soda solution sparging systems. This design was developed after discussions with and recommendations from C Manushewitz (1991).The proposed devices are in fact similar to the type of sparger which can be used for gas sparging in single equilibrium stage gas-liquid mass transfer operations in "bubble columns" or
in stirred tanks as described for example in the well known book by Treybal (1981). It is known that the Scandinavian
firm of Perstorp AB operate a reactor having what seems to
be an optimised sparging arrangement (Manushewitz, 1972) .
A cost estimate was recently prepared for manufacture and
installation of the sparging devices for the NCP reactor 68 (Cramer, 1992) . The amount came to R5 500 excluding VAT and
other costs related to "downtime" losses. Such a trivial sum might well be spent on what would in other words be a
plant scale experimental exercise in process optimisation.
8.2 THE DESIGN OF REAGENT SPARGING EQUIPMENT FOR THE NCP REACTOR
Short descriptions of two proposed devices, an acetaldehyde
sparger and a caustic soda solution sparger are given in
the following paragraphs. Calculations are shown and sketches showing the principal dimensions are given.
8.2.1 ACETALDEHYDE SPARGER
This reagent should be introduced in the reactor via a
sparger in the form of a ring conduit concentric with and
approximately at the level of or slightly above the lower
stage Rushton type turbine. This ring can have a smaller
clearance from the agitator than the present dip pipe
device; 150 mm or even 100 mm should be sUfficient. The
acetaldehyde reagent should be delivered through a number
of holes drilled through the ring conduit wall. These holes
should be of a suitable diameter to ensure a certain degree
of linear fluid velocity and turbulence as the reagent
enters the reactor bulk fluid. They should be spaced to
give maximum reagent distribution in the bulk space, and
positioned so that the impinging streams can be quickly and
thoroughly blended into the turbulent flowing streams of 69 reaction liquid emanating from the agitation impeller. With
a sparge ring level slightly above that of the agitator
downwards facing holes would be quite suitable. Figure 8.2
shows the proposed arrangement and dimensions are given in
the table as for figure 8.1. The design calculations are
shown in the following section.
8.2.1.1 calculations Acetaldehyde sparger
Number of holes 20
Hole Diameter 3 mm
Acetaldehyde liquid density 783 kg mo.'
Acetaldehyde viscosity 0.00025 Pa s
Quantity of reagent charged 950 kg
Time taken to deliver charge to reactor 80 minutes
Then Flow Rate
and orifice velocity
- 1t Flowrate+ (0. 003) ~*-=1.79ms- 1 4 70 And Orifice Reynolds Number
783*1.79*0.003=16800 0.00025
Reagent quantity charged, possibly, for an optimised reactor with a low Formaldehyde/Acetaldehyde Mol Ratio.
1235 kg
Charge delivery time for possible optimised reactor
60 min.
Then velocity
= 3.10 m S·I
And Reo = 29100 71 This device will thus sparge acetaldehyde effectively but not at unrealistically excessive velocities.
Pressure loss across the holes has been estimated using equation 5.19 and figure 5.12 from "Chemical Engineering", Volume 1 by Coulson and Richardson(1977). This method is
given for determination of flow from pressure drop in orifice meters and should be suitable for arriving at
values for flow through small holes, as in sparging
devices. The equation and graph are reproduced below.
100 I I I I 11'11 I • _ Sharp - edgea onflce I I diameter ratio '"' -' .... -, I .090 - OrIfIce dlam I (j~ ...... InSldl! pl=,e Ola% 01~ - I /' ...... OJ(,· -... '\\ \\ ~tg60 <, r-, V_ ~Q5q-f-. <,i'- ~L\ 070 so :::--,..,. -o I - .... l/, -, ~ h 30/ ~ 060 - ":I 02U' -u I Q) 050 -o u I 040
030 2 .. 10 20 40 60 10 10 Reynolds number through orifice 72
A suitable value for the discharge coefficient, Cll would be 0.65. Then using the rearranged equation 5.19 the head,
h, = 10 mm, which is equivalent to about 7.7*10.2 kPa or 0.01 psi. This value is quite low so one could easily allow for
fewer holes and higher velocities.
8.2.2 CAUSTIC SODA SOLUTION SPARGER
The physical properties of this liquid are very different
from those of the acetaldehyde. Also the caustic is fed
into the reactor by gravity flow through a control valve,
whereas the acetaldehyde is under a pressure of250 kPa in
the charge vessel. It will be possible to achieve linear
fluid velocity into the surrounding reactor fluid but in
terms of dimensionless Reynolds Number the degree of
turbulence will be low. This should not be a serious
disadvantage since the sparge ring can be positioned, like
the acetaldehyde device, so that the caustic can blend into
the turbulent streams emanating from the impeller region.
The caustic sparger will be similar to the acetaldehyde
sparger, but will be positioned at the upper agitation
stage and the number and size of holes will be different.
The device is also shown in figure 8.2 73
8.2.2.1 calculations caustic sparger
Number of holes 20
Hole diameter 5 mm
Liquid density 1500 kg m'
Viscosity @ 20°C 0.1 Pa s
Allow for a possible charge rate of 2500 kg of 50 % caustic
solution over a 60 minute period. i.e. :
and Reo:
1500*1.18*0.005=89 0.1
Here it should be noted that preliminary calculations have
indicated a drain time for the caustic charge vessel, with
control valve fully open, of about two minutes. 74
DIP PIPES \
L
I Ddp .1
Figure 8.1 : Stirred tank reactor showing present reagent -dip tubes'. 75
ACETALDEHYDE N (rpm) CAUSTIC -, /
B 50 Dsh
z en u
u
Of
Figure 8.2 : Stirred tank reactor showing proposed ring spargers ( Dimensions given overleaf) 76
Penta primary reactor with Improved Spargers
Principal dimensions
B Baffle width 0.25 em] C
Agitator turbine off-bottom distance 1. 37 (m]
Baffle distance 0.1 [m]
Interstage distance 1.5 em]
Agitator impeller diameter 1.2 [m]
Dip pipe distance (radial) 2.0 [m]
Shaft diameter 0.18 [m]
D. Proposed sparger diameter 1.5 [m]
DT Vessel diameter 3.2 em]
Dip tube height (of holes) 2.3 [m]
Liquid height 3.62 [m]
Blade length 0.3 em]
Shaft speed (65/60) = 1.083 [ S·I ]
Agitator Turbine blade width 0.25 em]
Number of agitation stages 2 [Number]
Sparger holes Acetaldehyde 20 (j) 3 mm
Caustic 20 (j) 5 mm 77
8.3 ALTERNATIVE SPARGING METHODS
An alternative method for reagent sparging is the use of a tee mixer in the cooling recirculation system as suggested
by LA van Niel. This would effectively change the reactor
configuration from a stirred tank with recirculation
arrangement to a system having a pipe reactor with cooler
and surge tank. An added advantage could be that the tank agitator be superfluous and switched off or dismantled
altogether.
The design of tee mixers is discussed in the following
chapter. The optimisation of a system for the NCP reactor
specifically is also dealt with. For devices of this type to be effective certain minimum entry-jet fluid velocities, relative to the main pipeline velocities, are necessary.
For the special case of the NCP plant such velocities could
only be achieved using a quite small diameter jet. The
pressure drop across an orifice of such small size (ca. 5mm
diameter) would be excessive, so much so that the 1ine
pressure of the Acetaldehyde feed system would be
insufficient to operate the mixer. Unless some other
arrangement regarding acetaldehyde delivery could be brought about, e.g. installation of a positive displacement
pump, it will be practically difficult to install a tee
mixer of other than a compromised design. Similar
arguments can be put forward concerning the use of high
velocity jet mixing in the stirred tank. 78
8.4 CONCLUSION: A RECOMMENDATION
It is strongly recommended that the NCP plant be modified
according to the concepts outlined above. The spargers
could be installed for a modest sum in relation to the
possible benefits. Effective optimisation can then be
achieved progressively by production trials. 79
FOOTNOTE TO CHAPTER 8 SPARGERS IMPLEMENTATION
The experimental details and results given in Chapters 6 and 7, and the recommendations for the design of sparging equipment were presented to National Chemical Products management in separate reports entitled "Pentaerythritol
Formation Reaction Laboratory Testwork with Agitation
Variation" and "Penta Reactor Improvement" during 1991 and
1992. The results were initially not thought to be
convincingly significant for the motivation of a plant
modification project. Subsequently, however, following
discussions with consultants of international standing,
partly as a result of the current emergence of South Africa
from a previously somewhat isolated situation with respect to industrial expertise and technology transfer, it became
apparent that the advantage to be gained from a reactor
improvement of this type could be considerable. The
proposal was thereupon reconsidered and at the time of the
present writing a reactor improvement project incorporating
the sparging equipment as described above is in an advanced
stage of implementation on the NCP plant. 80
CHAPTER 9 : TEE MIXERS
This chapter consists of a more detailed treatment of the mixing Tee as an alternative reagent sparging device. The first section is an introduction to the topic based on information from recent literature. The second discusses the development and technical feasibility for the
application of this type of device on the NCP reactor.
9.1 TEE MIXERS : GENERAL.
The design of Tee Mixers for fast reactions has recently
been dealt with in a paper by Forney and Gray (1990). The methods are based mainly on the experimental work of
Cozewith and Busko (1989) and other previous work notably
that of O'Leary and Forney (1985). The discussion given here is generally based on this approach.
One must assume that the fluids to be mixed are of the same
phase and fully mixable. A requirement is that fully
developed turbulent flow exists in both the pipeline and
the Tee inlet. The authors also stipulate impingement of
the jet on the opposite wall of the pipe.
The equations developed in the paper of Forney and Gray yield principally a velocity ratio for optimal mixing and 81 a -minimum pipeline distance for impingement of the jet
trajectory against the opposite pipe wall for the optimum
velocity ratio. Using correlations such as equation 5.19
and fig 5.12 from Coulson and Richardson, "Chemical
Engineering", Volume 1, as reproduced and used in the
previous chapter, one can estimate the pressure drop across the jet orifice.
9.2 A TEE MIXER FOR THE NCP REACTOR
The Acetaldehyde system at the Germiston plant functions
under pressure of approximately 250 kPa(g). It would be inadvisable to have a pressure loss across orifice much in
excess of say 5 kPa or 650 mm H20. From "Chemical Engineering", Equation 5.19, rearranged
G Ao = CoPJ2gh,)
Extracting from the graph, figure 5.12 in the same work,
the typical value 0.65 for Cu and the other values as in the nomenclature ,
Ao = 1.0891 * 10~ m2
Then d =fA" *n =0.0118 1ll
or d = 12mm 82
0:012 Then d = = 0.048 D 0.25
G 2.338 Then u I v = R = = = 1.47 pAu 1. 59
Then Re = pud = 84000, .. 1J.
i.e. flow is well into turbulent regime.
Now from Equation 6 of Forney & Gray (1990) the optimum u/v or R.
1 R = •••••••• (6) d ({3 _ d) DD
From the narrative of the above mentioned work a suitable value for (3 would be 1.6
Then R (Opt) = 13.42, i.e. the optimum value exceeds the real value by a factor of about 9.
The R function as described in the paper is then about 0.54 from the equation 9.
f(R) = 0.17 ..•...... (9) 0.1'" 0.35/Ru 5 83 Using equation 10 one can obtain the impingement distance, 1
L :: D
In this instance 1 is about 12 m, which is entirely unsuitable.
If on the other hand one could arrange for R to be at the optimum value, i.e. u = 13.42 * v = 21.3 m~1 then
1 = O. 17m. However from the Coulson and Richardson equation the differential pressure across the orifice is now of the order of 54 metres as static head, or about 540 kPa. Further calculations show that for delivery of
Acetaldehyde at the presently required rate an optimised jet would have a diameter of about 1.6mm, the velocity ratio would be about 120, the impingement distance 0.14m and the differential pressure of the order of 4300m or
33 000 kPa. In other words, the pressure drop would be
considerable although the velocity would still be well within the subsonic range. Installation may require an
augmented transfer system, such as a positive displacement
pump. The implementation of an optimised Tee mixer design
in the NCP plant would thus seem to be practically not feasible at present. Similar considerations should be
taken into account concerning the possibility of near-sonic
velocity jet mixing in the stirred tank, a technique
suggested by Dr Richter(1991} of Witwatersrand Technikon. 84
In conclusion it has been shown in this chapter that appropriate engineering methods have been developed and can be used for the design of an optimised mixing Tee and this type of device should be quite suitable for application to the pentaerythritol formation reaction. However in the special case of the NCP reactor an implementation by retrofit of this type of technology is probably not entirely feasible.
.: ! I 0 0 y ) • z ( I I I I I i • I x Pipeline i i : ! Side tee I Q IU ! i ! I r I " I
Figure 9.1 Tee Mixer 85
CHAPTER 10 STEADY-STATE FLOW REACTORS
Based on the results and discussions of the previous pages
this final chapter develops a general method for the design of continuous reactors for the industrial preparation of pentaerythritol. Broad principles are proposed but detailed design and calculations have not been given as this is not
considered as being within the scope of the present work. A more detailed treatment can be a topic for the future.
The first section is a general introduction to the topic
and a summing-up of relevant material from previous chapters. section 10.2 discusses the essential requirements
for the process design of a continuous reactor. The
conceptual design of a continuous pentaerythritol reactor is given in section 10.3. Finally in section 10.4 some
recent research is discussed and indications are given for
possible future work.
10.1 STEADY-STATE REACTORS GENERAL
Continuous reaction processes for Pentaerythritol
production have been mentioned in the literature, for
example in the monograph of Berlow et al. (1957) but
detailed descriptions are not generally available. This is
probably as a result of the confidential or proprietary 86 nature of the processes. The degree of chemical engineering design sophistication of these reactors is thus also riot known. In the following paragraphs a tentative design proposal for a continuous reactor, incorporating an optimised reagent sparging arrangement, will be developed.
As shown earlier in chapter 3, chemical reactor behaviour
for a process in which a fast reaction occurs between separate entering reactant streams is controlled by the
reactor contacting or flow patterns and chemical kinetics
becomes unimportant. contributions to the relevant flow
patterns are made by a.combination of interrelated factors,
these being the residence time distribution (R.T.D.), of material flowing through the vessel, the state of aggregation of the the material, and the earliness or
lateness of mixing of this material in the vessel. One
example is the combustion of fuel gas. The Pentaerythritol
condensation reaction is also just such a process and the
reactor can be treated accordingly. The reaction fluids and
more specifically the liquid Acetaldehyde reagent can
moreover be considered as microfluids rather than as
macrofluids; i.e. there is little or no tendency for the
occurence of aggregation or clumping of molecules or
molecular groups of the various chemical species and the
fluids tend to be fully miscible with and within one
another. This applies equally for gaseous as for liquid
acetaldehyde since it is quite readily soluble in the
aqueous reaction liquors. For this reason the use of a Tee 87
mixer type of reagent spa~ging device as described in the previous chapter cari be regarded as quite suitable for a
Penta reactor.
10.2 REACTORS FOR PENTA ESSENTIAL REQUIREMENTS
From all of the previous discussions and according to the indications of the experimental results, given in chapter
7, it can be seen that a penta production reactor should essentially have certain characteristics. These are a
turbulently f lowing stream of formaldehyde solution, an
effectively well admixed (into the formaldehyde) caustic
soda solution stream, and a likewise well admixed
acetaldehyde reagent stream. These requirements apply for batch reactors as well as in the case of continuous
processes. Furthermore, since it is desirable to keep the
relative concentration of acetaldehyde at a practically low
level, the reagent addition should be incremental as in
the semi-batch reactor and the reagent effectively
dispersed or sparged into the formaldehyde solution.
10.3 A CONTINUOUS PENTAERYTHRITOL REACTOR
10.3.1 CONCEPTUAL DESIGN
A first approach to the conceptual design of a continuous
Penta reactor can be achieved by considering the logical
extension of the semi batch CFSTR to a mUlti-stage series 88 of CFSTRs as in the Figure 10.1. In each successive tank in the series the incremental portion of acetaldehyde added undergoes complete reaction. This is analogous to describing the progress OT the proven stirred tank semi batch process in space and time instead of in time only.
The concept of an "ideal" continuous Penta reactor design with "maximal mixing " in a tubular type of device using a series of optimum Tee Mixers follows logically from the multi-stage CFSTR series idea. The arrangement would be as shown diagrammatically in Figure 10.2. 89
stage 1 stage 2 stage n
Reagent B Reagent B at ream 1 atream n
R·······lc.o \ \ \~' -, / '~
Figure 10.1: n Stirred tank reactor stages in series.
Reagent B: e.g.Acetaldehyde
! i I I I I I I I I
Reagent A Producta I . e.g Formaldehyde
...... -.._._ ...• Reagent C : e.g. Cauatlc
Figure 10.2 : • Side Feed Flow· steady-state reactor. 90
10.3.2 DISCUSSION
This type of reactor, the "Side feed flow reactor" is mentioned in paragraph 5.8 of the "Omnibook" (Levenspiel,1979), which also goes on to indicate the suitability of this flow pattern for reactions where the concentration of one of the reagents has to be kept low. Once again the ideal situation would be that each incremental addition of reagent undergoes complete conversion prior to addition of the succeeding increment. The overall fluid composition thus changes from point to point along the flow path as in the ideal plug flow pattern. In practice such a reactor could be designed to
allow for interstage mixing in static mixers positioned between the Tee mixers. A recently published study by Bourne et al. (1992) gives guidelines for the application of static mixers in chemical reactors. Temperature control in the continuous reactor could be engineered by incorporating a suitable counterflow cooling arrangement. 91
10.4 RECENT RESEARCH AND INDICATIONS FOR THE FUTURE
A patent recently mentioned in private communication to the author by Dr. C. D. Simon (Patent US2,818,443, Dec 1957; CA7347, 1958 discusses the idea of production of polyols by using anion exchange resins for the aldol reaction and hydrogenation over Raney Nickel for the reduction step. The problem with this route is that reported yields were low and hydrogen is an expensive raw material. Dr. Simon feels that a viable process could be developed if the yield can be optimised and another reduction process, specifically an electrochemical method could be used. The process would also be continuous. It would seem that the design of a commercial scale reactor of this type, especially in the case of fast chemical reactions, should also require, as in the case of the proposals outlined above, a considerable degree of optimisation as regards fluid flow and reagent mixing . In other words similar chemical reaction engineering principals apply in the case of heterogeneous reactions with solid catalysts or ion exchange resins.
In conclusion it is to be hoped that these few paragraphs have served to outline general methods of approach to the problems relating to the design of continuous pentaerythritol reactors. Some work has been done in the past and also more recently, but a detailed treatment of this topic can be the sUbject of some future work. 92
APPENDIX 1 : REACTOR PARAMETERS
The purpose of this appendix is to give a brief explanatory
discussion of the reactor agitation parameters and
coefficients used in the various chapters. The symbols are
as given in the nomenclature.
The feature articles by Bowen (1985,1986) represent a good
introduction and contribution to the understanding of
fluid mechanics involved in shear-sensitive liquid mixing.
The shear-sensitive applications are generally handled by
radial flow turbines such as the flat-blade turbine or flat
blade disc turbine, also known as the "Rushton" turbine.
The N.C.P. Penta reactor is equipped with two stages of the
latter type and Bowen's method can be used to estimate the
agitator discharge coefficient and discharge flow for the
system. A more complete account of the derivations is given
in the references. A brief summary will be given here.
Al. 1 DISCHARGE FLOW NUMBER, N,.
w. D(WID) \ \
Vmu = kNO(Dl1t 93
The diagram depicts the definition of geometry for the calculation. The discharge flow is obtained by integrating the parabolic radial velocity profile existing at the impeller tip over the width of the blade, and multiplying this area by the turbine circumference :
= 2nD \"\\'/2 q rad A ." vdz
Bowen shows that for the centreline radial velocity
1 He goes on to show that qr",i = 6.2ND/(W/O:\) (O.... /OT) II
J and by dividing both sides by ND A the flow number
Then Discharge flow, q"",= N'INO./Z
Where Z is the number of turbine stages 94
Al.2 AGITATION INTENSITY NUMBER, N1
This is given in the reference as being a function of bulk
fluid velocity, i.e.
NI = v,/ 0 . 0348
where v h is the bulk fluid velocity.
v, = 4NQNDA (D:\/D T ) 1/ 7T The Agitation Intensity Number is dimensionless and can
have a value on a scale from 1/3 to 10.
Al.3 BLEND TIME, Olt-
This is also dealt with in the article by Bowen (1985).
Tank turnover rate (Q/V) multiplied by 011 yields a
dimensionless group.
Bowen shows further that
01l(Q/V) = K = 0.05486 0IlN1/(V(IJ(Hr./DT)ln)
where the coefficient 0.05486 has the dimensions of
velocity, i.e. m/s.
Because the turns required for blending remain constant:
I 1 011 = KV ,.\ (Hr./ DT ) 1. / Nt
Thus for a given Agitation Intensity K/Oll = Q/V may be found. The reference gives graphs for correction of Q/V,
the Tank turnover rate for different values of HI./DT. 95
Al.4 POWER NUMBER, N p
For a 6 bladed Rushton turbine literature gives typically
4.6. Then for 4 bladed machines: Np = 4.6*4/6
For 2 stages Np = 2*4.6*4/6 = 6.2
5 Power drawn, P = Np P N'D
Torque, TA = P/2nN
Al.5 REYNOLDS NUMBER, Refllll
Dimensionless Reynolds Number of mixing for a baffled
stirred tank :
Refill) = pND" 1 / J.L 96
APPENDIX 2 CALCULATION OF RESULTS FROM EXPERIMENTAL DATA
This appendix gives an explanation of the manner in which
the raw data obtained from the laboratory experiments was
used to derive the results and produce the graphs shown in
chapter 7.
The chemical analyses done by R&D Analytical Laboratory
were reported as % m/rn in the reaction liquor. Further
data processing was carried out as follows :
A2.1 PRODUCTS AS % m/m OF TOTAL PRODUCTS
Mono-Penta % m/m = M * 100
M + D + B + U
where M = Mono-Penta % m/m reported
D = Di-Penta % m/m reported
B = BPMF % m/m reported
U = "Unknowns" % m/m reported
Di-Penta % m/m = D * 100
M + D + B + U
Similar formulae apply for the other species. 97
A2.2 YIELDS BASED ON ACETALDEHYDE
Mono-Penta Yield = M * R * 44 A* 136 * 100
Di-Penta Yield = D * R * 44 A * 254 * 100
BPMF Yield = B * R * 44 A * 290 * 100
Where A = Total mass.of Acetaldehyde in gm and the other symbols are as before. The numerical values in the
formulae are the mole masses of the species.
R, the total reaction mass was derived from the volumes of
reagents used in the experiments and the respective liquid
densities. The Formaldehyde solution density was estimated
using the equation and coefficients given in table 3 - 117,
"Densities of Aqueous solutions of Miscellaneous Organic
compounds" in Perry (1985). Sodium Hydroxide Solution
density was estimated using Table 3 - 90, also from Perry.
The density of pure Acetaldehyde at - 70°C was determined
by experiment in the laboratory, i.e. a container with the
liquid was cooled by immersion in a mixture of dry ice and
acetone and a measured volume weighed accurately.
These formulae thus give yields in moles of product per
mole of Acetaldehyde, the limiting reagent. 98
A2.3 SELECTIVITIES
The selectivities for the products were also calculated as
molar values.
Thus for Mono-Penta, S = (M * R) /136 (D * R)/254 + (B * R)/290
The calculations were performed wi t.h the aid of a Personal
Computer, using the Lotus 123 spreadsheet. The graphs
given in chapter 7 were constructed using the Harvard
Graphics package.
A2.4 EXPERIMENTAL RESULTS
Pertinent data, chemical analyses of the samples from the
experiments, and results as computed in the spreadsheet are
given in the following three pages. 99
PENTA REACTOR PARAMETER TESTS REACTION MASS PENTA TRIAL TOTAL FINAL MONO ERYTHROSE 01 BPMF UNKNOWNS gm. gm. %m/m %m/m %m/m %m/m %m/m
I'lR 5.::3 W M R. 1160 1065 6.9700 0.0000 1.1800 0.2900 0.0000 '1R 5.::3 W M R. 1160 1065 6.1100 0.0000 1.0700 0.0900 MR 5.::3 W M R. 1160 1149 7.5800 0.0000 0.6900 0.4200 0.2000 I'lR 5.3 WM R. 1160 1149 7.8500 0.0000 0.5700 0.5400 0.3000 MeAn results 7.1275 0.0000 0.8775 0.3350 0.1667 MR 5.3 NW M 1065 1054 6.8900 0.0000 0.4700 0.3200 0.3400 MR 5.3 NWM 1065 10S4 6.8800 0.0000 0.3700 0.3200 1).2400 MR 5.3 NWM 111:>0 1149 7.7000 0.0000 0.5900 0.49(11) 0.3600 MR S.3 NWM i reo 1149 7.4300 0.0000 0.4500 1).4800 0.2100 Mean results 0.0000 0.4700 "IR 6.5 WM R. 1::33 1148 b.1800 0.0000 0.7200 0.3000 MR 6.5 W M R. 1148 5.8500 0.0000 0.5200 0.2800 rlR 6.5 WM R. 7.0400 0.0000 0.3600 0.4300 0.1600 "lR 6.5 W M R. 1 ....,..,""" 6.9600 --- 0.0000 0.3400 0.4300 0.1800 MeAn results b.~075 0.0000 0.4850 0.3600 0.1700 MR b.5 NW M 1233 1148 6.3500 0.0000 0.3100 0.4000 MR 6.5 NWM 1233 1148 b.4500 0.3700 0.2900 0.2000 MR 6.5 NWM 1_.. 1266 ""- 6.9100 0.0000 0.2500 0.3600 MR b.S NWM 1266 6.0500 0.0000 0.2::00 0.2900 0.2800 MR 6.5 NWM 1233 1--- 6.2000 0.0000 0.::300 0.3::00 0.2300 MR 6.5 NWM 1233 6.8900 0.0000 0.4200 0.3800 0.1400 Me.n results 6.4800 0.0000 0.2860 0.3500 0.1760 "lR 9.0 WM R. 1684 1673 4.8300 0.0000 0.0800 0.1800 0.2100 MR 9.0 W M R. 1684 1673 5.1300 0.0000 0.1000 0.2100 0.1500 l'\R 9.0 W M R. 1684 1673 5.3000 0.0000 0.1100 0.2400 0.1400 5.0867 0.0000 0.0967 0.2100 0.16b7 MR 9.0 N W M. 1684 1673 4.8800 0.0000 0.0800 0.1900 0.1900 MR 9.0 N W M. 1684 1673 4.5800 0.0000 0.1100 0.2000 0.1700 MR 9.0 NW M. 1684 1673 2.1800 0.3500 0.0000 0.0400 0.0800 Mean results 4.7300 0.0000 0.0950 0.1950 0.1800 100
r TRIAL UNKNOWNS I'IONO P.-ERYTHR. D1 aPtotF UNKN. SlIl/1II ,., ,., ,., ,., ,.,
I'IR S.3 W 11 R. 0.0000 82.:5829 0.0000 13.9810 3.4360 0.0000 t1R :5.3 W 11 R. 84.0440 ·0.0000 14.7180 1.2380 0.0000
"IR S.3 W 11 R. 0.2000 8S.2643 0.0000 7.761S 4.7244 2.2497
I'IR :5.3 W 1'1 R. 0.3000 84.7732 0.0000 6.1SSS S.8315 3.2397
l'Ie." result. 0.lOb7 84.1661 0.0000 10.6540 3.807:5 1.3724
I'IR 5.3 NW 1'1 0.3400 85.9102 0.0000 5.S603 3.9900 4.2394
-I'IR-:5;;' 1'4- tH"t" - -.- ~.-2400- . 98.0922-- -0.0000- -- -'4.·7375 . . -l!r0913 3.0730
I'IR 5.3 NW 1'1 0.3600 84.2451 0.0000 6.4:551 5.3611 3.9387
I'IR :5.3 NW 1'1 0.2100 86. 697S 0.0000 :5.2509 5.6009 :=.4504 l'Ie.n results 0.2875 86.2363 0.0000 5.5760 4.7623 3.4254 t1R 6.5 W 1'1 R. 8:5.S333 0.0000 10.0000 4.1667 0.0000
I'IR 6.S W 1'1 R. 87.9699 0.0000 7.819:5 4.2105 0.0000
I'IR 6.S W 1'1 R. 0.1600 88.1101 0.0000 4.:5056 5.3817 2.0025
t1R 6.S W 1'1 R. 0.1800 87.9899 0.0000 4.2984 5.4362 2.2756
l'Ie." results 0.1700 86.5071 0.0000 6.b:559 4.7988 1.0695
I'IR 6.S NW 1'1 89.9433 0.0000 4.3909 S.66S7 0.0000
I'IR 6.S N W 1'1 8S.2~3 S.0616 3.9672 2.7360 0.0000
I'IR 6.5 N W 1'1 0.2300 89.1613 0.0000 3.2258 4.6452 ~.9677
I'IR 6.5 NW M 0.2800 8S.4:503 0.0000 3.2164 4.2398 4.0936
I'IR 6.5 NW 1'I 0.2300 88.82:52 0.0000 3.2951 4.5845 3.2951
I'IR 6.5 NW 1'I 0.1400 87.9949 0.0000 5.3640 4.8531 1.7880
Me." r-.ult. 0.1760 8S.87S0 0.0000 3.8984 4.7977 2.4289
t1R 9.0 W 1'1 R. 0.2100 91.1321 0.0000 1.5094 3.3962 3.9623
I'IR 9.0 W 1'1 R. 0.1500 91.7710 0.0000 1.7889 3.7:567 2.6834
MR 9.0 W 1'1 R. 0.1400 91.:5371 0.0000 1.8998 4.1451 2.4180
...." result• 0.1067 91.4801 0.0000 1.7327 3.7660 3.0212 t'ft 9.0 N .. 1'1. 0.1900 91.38:58 0.0000 1.4981 3.SS81 3.5581
t'ft 9.0 N .. 1'1. 0.1700 90.:5138 0.0000 2.1739 3."26 3.3597
I'IR 9.0 N .. 1'1. O.oeoo 82.2642 13.207:5 0.0000 1.5094 3.0189 ...." r-.ult. 0.1800 90.9498 0.0000 1.8360 3.7SS3 3.4589
-..
102
REFERENCES
Belkin,D.I.1979. Study of the dependence of pentaerythritol and dipentaerythritol yield on starting conditions for the
condensation of formaldehyde and acetaldehyde. Zhurnal Prikl. Khim.(Leningrad). vol.52, nO.1, p.237-9 (Russ.)
Berlow,E.iBarth,R.H. & Snow,J.E. 1957. The Pentaerythritols. A.C.S. monograph series, Rheinhold.
Bourne,John R.i Lenzner,Joachim & Petrozzi,Sergio.1992 Micromixing in static Mixers: An Experimental study. Ind. Eng. Chem. Res., vol. 31, p. 1216 - 1222.
Bowen,Richard L. Jr.,1985. Agitation Intensity : Key to Scaling up Flow-sensitive Liquid Systems. Chemical Engineering, vol. 92, no. 6, p.159-168.
Bowen,Richard L. Jr.,1986. Unravelling the Mysteries of
Shear Sensitive Mixing Systems. Chemical Engineering,
vol.93, no. 11, p.55-63.
Coulson,J.M. and Richardson,J.F., with editorial assistance of JR Backhurst & JH Harker. 1977. Chemical Engineering Volume 1. 3rd Edition. OXford: Pergamon Press. 449 p. 103 Cozewith, Charles; Busko,Michael Jr. 1989. Design Correlations for Mixing Tees. Ind. Eng. Chem. Res., vol. 28, no. 10, p. 1521 - 1530.
Cramer f P. L.1992. 2 Proposed Spargers for Penta Reactor. NCP-PED Memorandum. Ref. PED M.4/92/T.49/2/PLC/ydup.
Dickey, David S. 1991. Succeed at stirred-Tank-Reactor Design. Chemical Engineering Progress, vol . 87, no. 12. p.22-31.
Fasano,J.B. & Penney,W.R. 1991. Cut Reaction By-products by r Proper Feed Blending. Chemical Engineering Progress, vol.87, no.12, p.46-52.
ForneY,L.J. & GraY,G.E. 1990. optimum Design of a Tee Mixer for Fast Reactions. AIChE Journal, vol. 36, no. 11. p. 1773 - 1776.
Gould,E.S. 1959. Mechanism and structure in organic Chemistry. New York: Holt. p.389.
Ingold, C. 1969. structure and mechanism in Organic
Chemistry. 2nd Edition. Cornell University Press. p.680.
Jennery,G.R. 1991~. Pentaerythritol Reactor: Plant trials with controlled temperature variation. NCP Tvl. Technical
Department report. Ref. 160/91/GRJ/pmw. 104
JennerY,G.R. 1991h • Pentaerythritol Reactor optimisation. NCP Tvl. Memorandum Ref. 20191jGRJjgp.
JennerY,G.R. 1991c • Penta Reactor Optimisation: Action Plan Laboratory Test Work. NCP Tvl. Memorandum. Ref. 094/91/GRJ/pmw.
Koudelka,L. 1984. Calculation of rate constants and quantitative description of nonfully defined reaction systems. Chem Zvesti. vol.38, nO.5, p.649-61.
Laue,H.A.H. 1984. ~Pentaerythritol A survey of the Chemistry of the pentaerythritol process presented at a "Teach-in" held in the R&D conference room on 1984 10 19. NCP R&D . file : P2/1 HAHL/hb.
O'Leary, C.D.; ForneY,L.J. 1985. Optimization of In-Line
Mixing at a 90° Tee. I. & E.C. Process Des. & Dev., vol.24, p.332. 105 Levenspiel, Octave. 1972. Chemical Reaction Engineering. 2nd edition. New York: Wiley. p.578.
Levenspiel, Octave. 1979. The Chemical Reactor Omnibook. Distributed by OSU Book Stores.
Manushewitz,C. 1972. File on Perstorp visit, NCP Techcnical Department.
Manushewitz,C. 1991. Personal communication. NCP, Projects and Engineering Department.
March,J. 1968. Advanced organic Chemistry : Reaction
Mechanisms and structures. p.692.
Mvelase, T. 1992. Development of a reproducible laboratory test rig for the pentaerythritol reactor. NCP R&D report number R2/92, File P2/1.
Perry, Robert H. & Green,Don. 1984. Perry's Chemical
Engineers Handbook. 6th Edition. New York: McGraw-Hill.
"Polyhydric alcohols" in Encyclopedia of Chemical
Technology edited by Kirk Othmer, 3rd Ed., Vol 1, pp 778
789 (1978).
Richter, K.. 1991. Personal Communication. Department of
Chemical Engineering, Technikon Witwatersrand. 106
Simon, C.D. 1991~. The effect of reaction temperature on formation of Pentaerythritol and its by products, and the development of a new analytical GC chromatographic technique, and the regeneration of HPLC columns. NCP R&D Report No. R23/91.
Simon,C.D. 1991b • Literature review on the Pentaerythritol Reactor Chemistry, Colour Removal Systems, Waste Stream Utilisation and Lime Reaction. NCP R&D Report No. R28/91.
Simon,c.O.; Comninos,H.; cornish,L.; Liebenberg,o.O.; Nilson,O.; Shaw,P. 1993. Quantitative gas chromatographic determination of pentaerythritol and its' by-products by direct oximation and then silylation of aqueous solutions. Journal of Chromatographic science. Feb.1993 (in press).
Stoltz, G.O. & Salemi, G. 1979~.Literature survey on
pentaerythritol. NCP R&D report41/79 in file
P2/4.2(10/05/1979)
Stoltz,G.O. and salemi,G.1979b • Simulation of the IMC Pentaerythritol Process in the Laboratory. NCP R&D Report No. 111/79 in file P2/4.2 (1979/11/06).
Sykes,P. 1965. A Guidebook to mechanism in organic
chemistry. 2nd Edition. Longmans. p.172. 107 Trevoy,L.W.; Myers,M.E. 1963. Pentaerythritol derivatives.
I Mechanism of formation of dipentaerythritol. Can. J.
Chem., vol.41, p.770-6.
Treybal,Robert E. 1981. Mass Transfer operations. 3rd
Edition. Singapore McGraw Hill. p , 153 . (International Student Edition).
Van Niel,L.A. 1991. Handwritten comments on copy of
reference: JennerY,G.R. 1991c.• NCP PED.
Vik,J.E. 1972. Intermediates involved in the synthesis of
pentaerythritol and related alcohols. II. Kinetics of the base catalyzed aldol condensation reactions of formaldehyde
with its nearest homologs.Acca Chem. Scand. vol. 26, no.
8, p.316S-72 (Eng.).