WW wwmn llll‘lll'llllllll'lllllllll’l LIBRARY Michigan State University This is to certify that the thesis entitled PRACTICAL FLUIDS FOR FOOD RHEOLOGY AND PROCESS CALIBRATION presented by MITCHELL HULL has been accepted towards fulfillment of the requirements for MASTER'S degree in FOOD SCIENCE @FI/rri— Major professor Date 0/7/97 0-7639 MS U i: an Afi‘irman've Action/Equal Opportunity Institution PLACE II RETURN BOX to remove this checkout from your record. TO AVOID FINES rerun on or before dd. duo. DATE DUE DATE DUE DATE DUE '35? 2 21993 l l L__, 7.! MSU Is An Affirmdlvo Action’Equnl Opportunity Institution PRACTICAL FLUIDS FOR FOOD RHEOLOGY AND PROCESS CALIBRATION by Mitchell Hull A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Food Science and Human Nutrition 1989 b054444 ABSTRACT PRACTICAL CALIBRATION FLUIDS FOR FOOD RHEOLOGY AND PROCESS CALIBRATION by Mitchell Hull Corn syrups were shown to have potential for use as Newtonian calibration fluids in some applications where established standard oils are not suitable. Applications include food industry pilot and plant calibration operations that require large volumes of test fluids and subsequent water-based cleaning. Laboratory rheological analyses using both steady and dynamic shear conditions demonstrated that corn syrup vis- cosities are as constant as those of standard calibration oils. Some syrups showed significant elasticity that may be due to their tendency to dry upon exposure to air during analysis. The viscosities of some of the syrups did not significantly change during a five month period. With careful attention to shelf-life and protection from air drying, corn syrups can be used in many food indus- try fluid equipment system calibration and evaluation opera- tions. To Marie and Carolyn iii ACKNOWLEDGMENTS Dr. James F. Steffe, my major professor, deserves my sincerest gratitude and thanks for his unending support, interest, and patience in this project. I am graciously indebted to my guidance committee members: Mr. Gordon Casey, Dr. Mark Uebersax, and Dr. Robert Ofoli. For their long support in pursuing my Masters Degree and using Company equipment and materials in researching this thesis, I am deeply indebted to my supervisors at the Kellogg Company: Dr. Sungsoo Lee, Dr. Veronica Hicks, and Dr. Victor Fulgoni III. Many thanks belong to Dr. Sidd Purkayastha of A.E. Staley Co. for furnishing the corn syrups. I owe special thanks for this thesis’ inspiration to Dr. Steffe and to Mr. John Marlatt of the Kellogg Company. iv LIST OF TABLES LIST OF FIGURES NOMENCLATURE 1. INTRODUCTION 1.1. 1.3. 0 Calibration 1.1.1. 1.1.2. TABLE OF CONTENTS in Viscometry . . . Laboratory Applications Process Applications . . Justification for Study . . . . Study Objectives THEORETICAL CONSIDERATIONS 2.1. LITERATURE REVIEW 3.1. 3.2. Description of Newtonian Fluid Behavior 2.1.1. 2.1.2. General Calibration 3.1.1. 3.1.2. 3.1.3. 3.1.4. 3.1.5. Steady Shear Conditions Dynamic Oscillatory Shear Condit Definition Purposes General Procedure for Calibration Statistics of Calibration Requirements of Calibration Materials Components of An Accuracy-Based Measurement System 10 10 10 10 10 11 12 12 Or 4. 5. 3. 3. 3.4. 3. 5. Accuracy and Precision in Viscometry Absolute and Relative Viscosity . . Viscometry Calibration Materials and Methods 3.5.1. Definitive Methods . . . . . 3.5.2 Primary Reference Material 3.5.3 Reference Measurement Methods 3.5.4. Secondary Reference Materials 3.5.4.1. Survey of Internationally Recog- nized Materials . . 3.5.4.2. Reference Materials in the United States 0 3.5.5. Rheometer Manufacturers’ Recommendations 3.5.5.1. Hanging Weights . . 3.5.5.2. Calibration Fluids Alternative Secondary Reference Calibration Materials 0 O O O O O O O O O O O 0 3.6.1. logical Properties . . . . . 3.6.2. Survey of Literature Materials 3.6.2.1. Pure Compounds . . 3.6.2.2. Solutions . . . . . 3.6.2.2.1. Sucrose 3.6.2.2.2. Polymers 3.6.2.2.3. MATERIALS 0 0 O O O O O O O 0 0 O O O 0 METHODS O O O O O O O O O O 0 O O O O O O 5.1. Temperature Considerations . . . . . . . . 5.2. Determining the Degree of Newtonian Behavior vi Corn Syrups Relationship of Fluid Structure to Rheo 17 19 19 21 21 24 25 26 28 28 29 3O 30 32 32 34 34 37 39 43 44 44 44 5.2.1. Linearity of Shear Stress vs. Shear Rate in Steady Shear . . . . . . . . . . . . 44 5.2.1.1. Haake Coaxial Cylinders . . . . 44 5.2.1.2. Carrimed Cone and Plate . . . . 48 5.2.2. Dynamic Oscillatory Shear Experiments . 50 5.2.3. Time-Dependency . . . . . . . . . . . . 51 5.2.3.1. Multiple Analyses of Same Sample Load 0 0 O O O O O O O 0 O O O 51 5.2.3.2. Separate Analyses Conducted Days or Months Apart on the Same Sam- ple O O O O O O 0 O O O O O O O 52 6. RESULTS AND DISCUSSION . . . . . . . . . . . . . . . 53 6.1. Haake Sensor Windup . . . . . . . . . . . . . . 53 6.2. Degree of Newtonian Behavior . . . . . . . . . 57 6.2.1. Steady Shear . . . . . . . . . . . . . . 57 6.2.1.1. Haake Coaxial Cylinders . . . . 57 6.2.1.2. Carrimed Cone and Plate . . . . 59 6.2.2. Dynamic Oscillation . . . . . . . . . . 65 6 o 3 0 Time and Shear‘DePendenCY o o o o o o o o o o o 67 6.3.1. Up—Ramps vs. Down-Ramps . . . . . . . . 67 6.3.2. Repeated Measures on Same Sample Load . 75 6.3.3. Repeated Measures on Different Sample Loads O O O O O O O O O 0 O O O O O O O 77 6.4. Sample Stability in Handling . . . . . . . . . 80 7. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . 84 8. SUGGESTIONS FOR FURTHER STUDY . . . . . . . . . . . 86 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . 88 APPENDIX 1. ORIGINAL HAAKE CONCENTRIC CYLINDERS DATA . 93 APPENDIX II. ORIGINAL CARRIMED STEADY SHEAR DATA . . . 102 vii APPENDIX III. ORIGINAL CARRIMED OSCILLATORY SHEAR DATA 109 viii Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. Table 8. Table 9. Table 10. Table 11. Table 12. Table 13. LIST OF TABLES Selected sucrose-water solution viscos- ities at 20°C. . Composition and physical properties of A.E. Staley Co. corn syrups. . . . Haake concentric cylinder dimensions. . . Linearity of shear stress - shear rate analysis results as expressed by corre- 2 lation coefficient, r , for low viscosi- ty fluids, concentric cylinders. . . . . Linearity of shear stress - shear rate analysis results as expressed by corre- lation coefficient, r“, for medium and high viscosity fluids, concentric cyl- inders. . . . . . Linearity of shear stress - shear rate analysis results as expressed by corre- 2 lation coeffecient, r , cone and plate. . Carrimed dynamic Cannon S8000 oil of 0.01 radians. Carrimed dynamic Cannon S8000 oil oscillation results for at a constant amplitude oscillation results for and Staley 3260 syrup. . Results of analyses repeated on the same load of various oils and syrups, concen- tric cylinders. . Results of analyses repeated on differ- ent sample loads of various oils and syrups, concentric cylinders. . . . . . . Original Haake Data . . . . . . . . . . . Original Carrimed steady shear data . . . Original Carrimed oscillatory data . . . ix 36 40 47 58 61 66 68 69 76 79 93 102 109 N Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure LIST OF FIGURES Simple shear model. . . . . . . . . . Theoretical shear stress - shear rate plot for a Newtonian fluid. . . . . . Carrimed cone and plate geometry with solvent trap. . . . . . . . . . . . . Haake "Windup" effect. . . . . . . . . Haake step-ramp program and sensor lag, Cannon 830000 oil. . . . . . . . Linearity of low viscosity fluids, concentric cylinders. . . . . . . . . Flow curves of lower viscosity fluids, cone and plate. . . . . . . . . . . . Flow curves of higher viscosity flu- ids, cone and plate. . . . . . . . . . Theoretical thixotropic loop. . . . . Flow curves of medium viscosity flu- ids, cone and plate. . . . . . . . . . Flow curves of high viscosity fluids, cone and plate. . . . . . . . . . . . Flow curves with thixotropic loop of low viscosity fluids, cone and plate. Flow curves of Staley Sweetose 4300 corn syrup, effect of solvent trap on cone and plate results. . . . . . . . 49 54 56 60 63 64 71 72 73 74 81 70 7m NOMENCLATURE area of ideal parallel plate, m2 force required to maintain velocity v of ideal par allel plate, N storage modulus, Pa time, s distance between ideal parallel plates, m angular dynamic strain, radians angular strain amplitude, radians -1 shear rate, 3 maximum shear rate attained during a steady shear analysis angle of the phase lag between the imposed stress and the resulting strain, degrees Newtonian viscosity, Pa 3 loss modulus or dynamic viscosity, Pa 3 angle of cone, degrees kinematic viscosity, m3/s ideal plate velocity, m/s density, kg/m2 shear stress, Pa shear amplitude, Pa yield stress, Pa angular frequency, radians/s xi Chapter 1 INTRODUCTION 1.1. Calibration in Viscometry 1.1.1. LaboratoryiApplications Calibration in Viscometry is carried out to ensure that the analytical method being calibrated provides accurate re- sults. Not only is calibration used in the design and manu- facture of instruments (e.g., Brookfield) that measure true results only for ideal fluids, calibration is also performed by purchasers of "absolute" viscometers to assure themselves that the equipment is operating as designed (Delaney, 1988). Calibration is the process of determining the response of a viscometer to known fluids. One, therefore, must have fluids whose viscosities are accurately known. The calibra- tion procedure should be easy to conduct, thus permitting a proper statistical determination of the error of the viscom- eter. 1.1.2. Process Applications The design and evaluation of process equipment that handle fluids is a continuing problem in food and other industries. Laboratory measurements of fluid flow behavior are sometimes used to design new equipment and to predict how a new fluid will behave in an existing fluid-handling system. Many process conditions cannot be accurately pre— dicted, however, due to limitations in mathematical 2 description of complex geometries such as those found in pumps, filters, manifolds, heat exchangers, and extruders. Process engineers therefore frequently prefer to eval- uate a fluid handling system without having to conduct pro- duction-usage tests. Pilot scale production systems can be evaluated with a known fluid to determine the likelihood of success with the intended fluid. This pilot or process "calibration" can also be used to confirm operation of in- line test transducers such as pressure sensors, thermocou— ples, and flowmeters. Finally, process calibration can help verify mathematical models of fluid flow in the process under study. These models are sometimes used to predict behavior of complex fluids in the process and therefore help guide manipulations of the process to better attain produc- tion objectives. 1.2. Justification for Study Commercially available, certified viscosity standards are not suitable for many viscometer calibration needs, particularly those in the food industry. Costing $30-$40 per pint (Cannon Instrument Co., 1988), they are practical only for small volume viscometers found in the laboratory. All require organic solvents for cleaning, the higher vis- cosity standards being readily soluble only in halogenated hydrocarbons such as chloroform. Many applications in the food industry requiring cali- bration fluids are not well served by the current recognized standard fluids. A partial list of applications includes 3 laboratory viscometers requiring substantial sample volume (e.g., some Brookfield instruments), tube viscometers beyond capillary size, process viscometers (Cheng et al., 1985), mixers, pumping equipment, manifolds, filters and screens, most heat exchangers, and extruders. There exists a need, therefore, for calibration fluids which are inexpensive, well-characterized, reasonably ideal in behavior and stability, can be easily cleaned with water, and are suitable for use in food processing. 1.3. Study Objectives There are three basic objectives of this research: 1. Identify suitable materials for food industry use. A fluid suitable for use in the food industry should have two main characteristics: A. Ideal flow behavior. A calibration fluid should conform to a simple model of behavior when undergoing a shear stress or strain. The simplest model is that of Newton, who described the force of shearing a fluid as directly proportional to the rate of shear. A so-called Newtonian fluid is not time-dependent and has a constant viscosity at any rate of shear. B. Ease of use. A calibration fluid should be easy to use, to encourage a frequency of testing that is statistically significant. "Easy to use" ideally means that the fluid should be clean- able with water, safe, simple to prepare or ready for use as 4 purchased, stable during storage and use, easily disposable, and low in cost. 2. Determine extent of Newtonian behavior of suitable fluids. Viscometric analyses of a few fluids, determined to be suitable from the above survey, will be presented. The analyses will focus on the constancy of viscosity over a range of shear rates (experimental conditions). Comparisons will be made between currently accepted standards and poten- tially more practical alternatives. 3. Recommend fluids and procedures. Fluids will be recommended, based on experimental evi- dence. Procedures for handling and storage will also be suggested for specific fluids. Chapter 2 THEORETICAL CONSIDERATIONS 2.1. Description of Newtonian Fluid Behavior The ideal Newtonian fluid will be described under steady and dynamic shear conditions. 2.1.1. Steady Shear Conditions Conditions of steady shear can be most easily described in terms of the simple shear model depicted in Figure 1 (Boger et al., 1985). A fluid is held between two parallel infinite planes separated by a distance y. The top plate of area A is moving to the right with a steady velocity, v. The fluid is sheared between the plates. The force required to maintain this velocity is F. This laminar flow field can be mathematically expressed by relating the force per unit area to the rate of shear: F/A = n - v/y [1] where n is a proportionality constant called viscosity. The shearing force per unit area is commonly called the shear stress and is denoted by 0. The rate of shear v/y is called the shear rate and is denoted by y. The above relationship is thus written: 0 = fl ' V [2] A Newtonian fluid has a linear relationship between shear stress and shear rate, passing through the origin of a plot of shear stress vs. shear rate, as shown in Figure 2. A Newtonian fluid has a constant viscosity at constant tem- perature. __§. :2» “v Figure 1. // /T // // Simple shear model. // ,/ /’ / // O W/ Figure 2. Theoretical shear stress - shear rate plot for a Newtonian fluid. 8 Some analytical instruments, notably capillary visco- meters, measure kinematic viscosity, u, directly: u = n / p [3] where p is the fluid density. 2.1.2. Dynamic Oacillatory Shear Conditions This flow mode is often used to measure the behavior of viscoelastic fluids which exhibit properties between ideal solids and ideal liquids. By using this flow mode one can determine if there is any residual solid-like behavior in a candidate calibration fluid. Consider the parallel plates discussed in section 2.1.1. Instead of imposing a steady linear velocity v on the upper plate, the plate is driven with a sinusoidally varying strain (distance the plate is moved) y at a frequency of w/2n, where o is the angular frequency (Whorlow, 1980). The shear stress, 0 = 00 - cos(ot) [4] produces a strain, V 70 - cos(ot-6) [5] where 5 is the angle of the phase lag between the imposed stress and the resulting strain, yo is the angular strain amplitude, and t is time. The storage modulus, G’ = 00 - cos(5) / 70 [6] gives the in-phase stress amplitude, while the dynamic vis- cosity is expressed as n’ = 00 - sin(6) / Yo [7] 9 The dynamic viscosity n’ approaches the steady viscosity n at very low frequencies. G’ and n’ are related to each other by tan (5) = wn’/G’ [8] The value of tan (6) should be infinite for an ideal pure fluid, since G’ << n’. A purely solid (Hookean) material would be strained exactly in phase (6 = 0), while a purely viscous fluid (New- tonian material) would be strained exactly 90' out of phase. This is because an imposed stress is directly proportional to the strain for a pure solid, whereas the stress is pro- portional to the rate of strain v for a pure fluid. From the above discussion, the behavior of a Newtonian fluid in dynamic oscillatory testing becomes clear. The experimentally determined value of G’ should be zero, while the value for n’ should be the same as the viscosity deter- mined in steady shear: n’ = R In fact, Boger et al. (1985) suggest that the accuracy of dynamic measurements can be checked with a Newtonian fluid of known viscosity. Chapter 3 LITERATURE REVIEW 3.1. General Calibration Considerations 3.1.1. Definition Analytical chemists understand calibration to mean the indexing of instrument or indicator response to the concen- tration of the analyte species of interest. Calibration may be thought of as relating a measured property of a sample to an estimate of the desired property of that sample (Werni- mont, 1985). 3.1.2. Purposes Calibration is done for one or both of the following rea- sons: a) when the analytical theory cannot readily predict quan- titative instrument response from analyte concentration; b) when significant errors arise during an analysis based on quantitatively fundamental theory. In practice, nearly all instrumental analyses use calibra- tions to assure the highest accuracy possible (Skoog and West, 1971). 3.1.3. General Procedure for Calibration Calibration in analytical chemistry is typically carried out as follows (Miller and Miller, 1984): The analyst takes a series of samples (normally at least three or four, and possibly several more) in which the concentration of the analyte is known. These calibration standards are measured in the analytical instrument under the same conditions as those subsequently used for the test (i.e., the "un- known") samples. Once the calibration graph has 10 11 been established the analyte concentration in any test sample can be obtained by interpolation. The calibration graph quoted above is a scatter (x,y) plot of the instrument responses to the known (calibration) samples versus the values assigned to the known samples. 3.1.4. Statisticaaof Calibration There are several questions about a calibration graph that must be addressed (Miller and Miller, 1984). These include: 1) is the calibration graph linear or curved; 2) what is the best line through these points; 3) what are the estimated errors and confidence limits for the regression parameters that describe the line; 4) when using the cali- bration to measure test samples, what are the errors and confidence limits for the determined value; and 5) what is the limit of detection of the method? The widely used statistical methods for computing a calibration regression are based on two assumptions: 1) the errors in the calibration graph occur only in the instrument response (y axis), i.e., the values assigned to the known calibration materials are assumed to have no error; and 2) the magnitude of the error in the instrument response is independent of the magnitude of the value assigned to the calibration material. Both assumptions are violated in most calibrations, but only the second has occasionally serious results. A detailed discussion of the statistics of cali- bration can be found in Miller and Miller (1984). 12 3.1.5. Reguirementaaof Calibration Materiala The most important characteristic a calibration material must have is credibility. The analyst must accept that the value assigned to the material accurately reflects the true value. The universally accepted strategy to attain this confidence is to obtain and properly use materials that are associated with a national standards organization program. The components of such a program are discussed in Section 3.2. Other requirements typically made in calibration mate- rials are: (1) independence from the method in question; (2) sufficient precision in the value ascribed to the mate- rial, so that none of the variability attributed to the instrument readings may be assumed to derive from the stan- dards; and (3) inclusion of interfering "matrix" effects (i.e., non-ideal behavior such as elasticity, temperature sensitivity, and multiple phases) that might be encountered in actual samples (Wernimont, 1985). 3.2. Componenta of An Accuracy-Based Measurement Syatam The "systems approach" to assuring measurement compati- bility, or accuracy, has been described (Uriano and Gravatt, 1977) as having six necessary technical components: 1. Basic Measurement Units These provide self-consistent measurement scales that can be related to each other in exact physico-chemical equations. The Systeme International d’Unites (SI) are recognized by most nations as the basic units. There are 13 seven base units in the SI system: mass, length, time, electric current, thermodynamic temperature, luminous inten- sity, and the quantity of substance (the mole). The SI units for viscosity and kinematic viscosity are Pa 3 ([N/mz] s) and mz/s, respectively. Commonly used units for viscos- ity are centipoise (cP), where 1 cP = 0.001 Pa 3, and for kinematic viscosity, centistokes (cSt), where 1 cSt = 0.0001 mz/s (ASTM, 1983). 2. Definitive Measurement Methods Definitive methods of chemical analysis are those that have a valid and well-described theoretical foundation, give negligible systematic experimental errors and have high levels of precision. These methods evaluate the property in question either directly, in terms of the fundamental units of measurement, or indirectly relate the property to the fundamental units via exact mathematical equations. Such methods give "true results" with high reliability. The measurements obtained from these methods are said to be absolute measurements. 3. Primary Reference Materials Reference materials are a class of well-characterized, stable, homogenous materials produced in quantity and having one or more physical or chemical properties experimentally determined within stated measurement uncertainties. Primary reference materials are those having properties certified by a recognized national standards laboratory. The measurement techniques used to certify primary reference materials are 14 the most accurate and reliable methods available: the definitive methods. The International Union of Pure and Applied Chemistry (IUPAC) and the International Organization for Standardization (sometimes referred to as the "Interna- tional Standards Organization" and abbreviated hereafter as 108) have recommended the use of the term "Certified Refer- ence Material" (CRM) for primary reference materials; in the U.S., the National Bureau of Standards (NBS) uses the term Standard Reference Materials (SRM). CRMs are produced in relatively small amounts to the highest standards of accura- cy, and are intended for use where accuracy is the most important requirement, regardless of cost. A primary use of CRMs is to serve as the foundation of accuracy for the development of reference methods. 4. Reference Measurment Methods A reference method is defined as "a method of proven and demonstrated accuracy " (Cali et al., 1975). Reference methods represent the primary mechanism by which the accura- cy of a definitive method or CRM is transferred into wide- spread use in the field. The reference method has proven accuracy, but may lack the direct or indirect relationship to fundamental units of the corresponding definitive method. The reference method, like the definitive method, may not be acceptable for routine field use when cost and/or speed are important. 15 5. Secondary Reference Materials While primary reference materials are usually produced by a national standards laboratory such as NBS, secondary reference materials are usually developed for use in day-to-day field operations where cost and moderate accuracy are the most important requirements. The materials are usually produced by commerical or individual laboratories, and are meant for direct use as working standards. 6. Field Methods These are routine methods typically used in normal, everyday operations requiring high throughput and low cost. Many of these methods are instrument-based and are based on comparative analytical principles, thus requiring the use of reference materials to correct for inherent systematic biases. 3.3. Accuracy and Precision in Viscometry The accuracy of rheological analyses can only be firmly established for Newtonian fluids. For fluids of unknown behavior, accuracy is more elusive (Prentice, 1984): The usual method of achieving accuracy is by cali- bration. To the classical physicist this may seem so axiomatic that it does not need to be stated. However, when dealing with rheologically interesting materials, the response of any instrument cannot be separated from the properties peculiar to the mate- rial. If a viscometer is calibrated, as it often is, by the use of a standard Newtonian fluid of known or predetermined viscosity, then its calibra- tion can hold only for Newtonian fluids. Ideally, for use with a non-Newtonian fluid, it would be calibrated with one of similar characteristics. This, though, requires a knowledge of the very properties that are being sought. It is a circular argument and there is no way of breaking into it. One must therefore have recourse to calibration in 16 terms of the fundamental units. The fundamental quantities which are measured by a rheometer are usually a torque...and a velocity. However, a knowledge of the torque and velocity are only a means to an end for the rheologist. In order to define the properties of his materials properly, he needs to establish the relationship between stress and strain, and he requires his result to be unin- fluenced by the instrumental means he uses to obtain it. Herein lies the problem of achieving accuracy. It has been shown that response of the instrument is conditional upon the properties of the sample mate- rial. Consequently, it is not possible to derive an unambiguous stress or strain.... Marvin (1972) also discusses this circular logic in viscosi- ty measurement: Stress, being defined in terms of forces acting across imaginary planes in a material, cannot be measured directly. It must be inferred from mea- surements made at a surface of the material, and its calculation implies some assumptions about the rheological properties of the material studied. One indicator of the comparative lack of accuracy in the field of rheology is the value assigned to the viscosity of water. The viscosity of pure water at 20.00'C is the inter- nationally agreed reference point for Viscometry. The most recent fundamental measurement of the viscosity of water compared two independent analyses (Marvin, 1971). The ac- curacy for each method was estimated at 0.1%, yet the re- sults of the two methods differed by 0.5%. While the uncer- tainty assigned to the value of the absolute reference was therefore declared to be t 0.25%, agreement between labora- tories is better than 0.1% in many cases (108, 1977). The latter precision figure is for capillary viscometers used under very stringent protocols. 17 The uncertainty assignable to the accuracy of the best capillary measurements of high viscosity fluids cannot be less than one percent (Marvin, 1971). Most field methods, made with instruments rather than with capillaries, have considerably larger precision errors. R.S. Marvin, a re- tired NBS rheologist, has commented that few instruments, when presented with two fluids differing in viscosity by five percent, will detect any difference (Manning, 1987). Another rheologist (Tung, 1987), working in food science, called rheology "the five percent science." 3.4. Absolute and Relative Viscosity Officials of NBS have written (Uriano and Gravatt, 1977) that An absolute measurement of a chemical property such as composition is generally made by a method (either instrumental or manual in nature) in which the property in question is either directly evaluated in terms of the fundamental units of measurement or indirectly related to the base units via exact mathematical equations. The above definition of an absolute method is restricted to very well-defined circumstances. Absolute viscosity measurements are those conducted with a complete understanding of the interaction of fluid and instrument. In the strict sense of Uriano and Gravatt, absolute measurements have been published only a few times, mostly of water. All other measurements have been made in explicit or implicit relation to water. More commonly in scientific literature the term "abso- lute viscosity" is used to mean "true" viscosity, as opposed 18 to "relative" or "apparent" viscosity. Absolute viscosity in this context is the relationship between shear stress and shear rate. A necessary condition in deriving an absolute viscosity is that the shear field developed in the measuring device (called an "absolute viscometer") is well understood. This means that both the shear stress and the shear rate must be known with reasonable accuracy. A measured absolute viscosity will be the same regardless of the type of instru- ment used in the measurement. An absolute viscosity mea- sured in a laboratory, for example, will therefore truly represent the viscosity of the same fluid in a defined process situation. Relative viscosity generally refers to those measure- ments made on fluids that are repeatable only when using a specific instrumental configuration and technique. Another characteristic typical of relative viscosity measurements is that the conditions of shear stress and/or shear rate are unknown for the analysis. A relative viscometer can be calibrated with suitable (Newtonian) reference fluids so that the relative viscometer can give "true" or absolute viscosity when measuring Newtonian fluids, whose viscosities do not depend on shear rate. An example of a relative viscosity analysis is one made using a Brookfield RV viscometer. The fluid-contacting sensors for this viscometer have complex geometries that make it difficult, or impossible, to accurately describe the shear field present in the fluid. The Brookfield’s results 19 can be compared with other identically obtained analyses to serve as a product development yardstick, for instance. The same results, however, cannot be used to predict fluid behavior under actual industrial processing conditions. A cause of some confusion is the apparent paradox that instruments such as the Brookfield can also be used to mea- sure the absolute viscosity of Newtonian fluids. This apparent contradiction is a consequence of the use of Newto- nian reference materials to calibrate the torque—measuring response of the instrument. Because these calibration standards do not depend on a knowledge of the rate of shear, the calibration curve is able to successfully relate the torque to the viscosity of materials whose viscosity does not vary with shear. 3.5. Viscometry Calibration Materia1§_and Methods This section will discuss how standards of accuracy in Viscometry are disseminated from the national standards laboratories to routine laboratory use. The components of this distribution system (units, analysis methods, and standard materials) were discussed in Section 3.2. The as- signment of specific materials and methods to the components is the author’s. 3.5.1. Definitive Methods No one method is recognized to be the most definitive, most accurate method for measuring viscosity. Most absolute measurements attempting an accuracy of better than one percent when measuring the viscosity of water (the 20 internationally recognized primary reference material for Viscometry) have been made using capillary flow (Marvin, 1971). Bingham and Jackson (1919) were the first to deter- mine an internationally accepted value for the viscosity of water by evaluating published capillary results dating from 1840. The currently accepted value for the viscosity of water was published by Swindells et al. (1952). The method used four different capillary viscometers, one pair having the same length but different radii, the other pair having the same radii but different lengths. The work was conducted over a twenty-year period. Between 1952 and 1971, only three attempts were made to measure the absolute viscosity of any liquid. Roscoe and Bainbridge (1958) suspended a water-filled glass sphere from a torsion wire, and measured decrement in the oscillation period. Maliarov (1959) measured pressure drops across two capillaries connected in series through a central water- -filled reservoir. Finally, Kawata, Kurase, and Yoshida (1969) used a capillary, essentially the same as that of Swindels et al. (1952), to measure the viscosity of a hydro- carbon liquid. Most recently, Marvin (1971) discussed absolute viscos- ity measurements made by colleagues at NBS on water using two independent methods. White and Kearsley (1971) used a torsional pendulum to measure the period of oscillation of a sphere through water. Penn and Kearsley (1971) measured 21 pressure drop through taps in the walls of a channel through which water was pumped. The channel was a novel design formed from two accurate cylinders and an optical flat, yielding a triangular cross-section with one side flat and the other two sides circular arcs. Both methods had an estimated accuracy of 0.1 percent, yet the two results differed by 0.5 percent. 3.5.2. Primary Reference Material IUPAC (Marsh, 1980) and IOS (1977) have stated that "the internationally recognized reference material is water in equilibrium with air at 293.15 K and atmospheric pressure," where the value of atmospheric pressure is 101.325 kPa. The recommended method of preparation cited by IUPAC is "purifi- cation by double distillation and passing through a fine sintered glass frit (to saturate the sample with air) just prior to use (Swindells et al., 1952)." IUPAC (Marsh, 1980) and IOS (1977) have accepted the recommendation of Marvin (1971): ...that viscometer calibrations be based on the following value for the viscosity of water = 0.001002 Pa 3 [at 293.15 K] with the proviso that when only agreement between measurements made in different laboratories is important, the uncertainty associated with this value be ignored but that when a true value of the viscosity is required, an uncer- tainty of $0.25 percent be assigned to this value (Marsh, 1980). 3.5.3. Reference Measurement Methoaa Recall that a reference method is defined as "a method of proven and demonstrated accuracy " (Cali et al., 1975). These methods are used to transfer the accuracy of a defini- 22 tive method or certified reference material (CRM) into wide- spread use in the field. In the U.S., the American Society for Testing Materials (ASTM) has specified the reference method for transferring the result of the absolute viscosity measurements of water into widespread field use. The method (ASTM, 1987) uses a series of "master viscometers" with different calibration constants to successively measure water and oil standards. The master viscometers are capillary instruments, either of the Cannon (1944) or Ubbelohde (1936) type, specially de- signed to minimize errors due to surface tension, kinetic energy, and capillary end effects. Capillary viscometers are apparently the universal choice for viscosity measurement where high precision is required. Temperature control is often the factor limiting the precision attainable in the measurement of viscosity. The percentage change of viscosity for many fluids typically ranges from about 2 percent per '0 at 0.001 Pa 3 to 10 percent per '0 at 10 Pa 3. For accurate viscosity measure- ment the temperature must therefore be controlled to 10.01°C or better. This level of control is most easily realized with a viscometer that can be completely immersed in a thermostatic bath. A capillary viscometer is one of the few viscometers easily usable in such a bath. The first part of the ASTM method uses two master visco- meters, referred to as "master class" instruments by IUPAC (Marsh, 1980), with calibration constants in the range of 23 0.001 to 0.003 cSt/s. Water under the primary reference material conditions at 20°C is used to determine the calib- raton constant of the two instruments. The calibrated master class viscometers are then used to measure two Newtonian oils (see Secondary Reference Materi- als, Section 3.5.4.) at 40°C. The first oil must have a kinematic viscosity of about 3 cSt at 40°C, the second a kinematic viscosity of about 8 to 9 cSt. The now-calibrated oils are then used to determine the calibration constant of a third master viscometer in the range of approximately 0.003 to 0.009 cSt/s corrected to 20°C. Additional master viscometers and additional viscosity oil standards can be calibrated by extending this "step-up" procedure until the desired viscosity range is achieved. Steps between successive viscosities or calibration con- stants must increase by a factor of three or less. Oils are calibrated at other temperatures using the average result from two master instruments. Corrections must be made, where applicable, for differences in acceleration due to gravity, temperature, buoyancy, density, and surface ten- sion. The repeatability (difference between repeated tests) of the method is such that the difference between successive test results should exceed 0.1% of their mean in less than 1 case out of 20. The reproducibility of the method (differ- ent operators and laboratories) is better than 0.35%, analo- gously defined. The accuracy of standards over 0.1 Pa 3 24 calibrated by this method should be assigned a value of i0.3%, the uncertainty arising from the error in the multi—step and in the limited stability of the oils (Marvin, 1972). Other nations use an essentially identical approach to that of the United States. According to Marsh (1980): ...reference materials with certified values are used to calibrate viscometers. The certified values of these reference materials are always determined by comparison with a viscosity of water, either directly or indirectly, through a chain of interme- diate reference liquids and master class instru- ments. The consistency of the various national viscosity scales is checked by a continuing pro- gramme of direct international comparisons made by the exhcange of both master viscometers and refer- ence materials between the various suppliers. The ASTM methods are apparently well-respected in the international community. As an example, consider the work of Bauer and Meerlander (1984). They used ASTM method D 2162 (1973) to measure water as the basis for an investiga- tion of low-viscosity calibration materials. See Section 3.6.2. for further discussion. The authors were with the Physikalisch-Technische Bundesanstalt, the national stan- dards laboratory of the Federal Republic of Germany. 3.5.4. Secondary Reference Materials IUPAC has published a survey of secondary reference materials: "Recommended Reference Materials for Realiz- ation of Physicochemical Properties. Section: Viscosity" (Marsh, 1980). The recommendation concerns only reference liquids which are Newtonian. One may infer that 25 non-Newtonian fluids are not recommended by IUPAC as viscos- ity reference materials. 3.5.4.1. Survey of Internationally Recognized Materiala The fluids from the 10 international suppliers cited in the IUPAC survey include petroleum oils, polyisobutenes, silicone oils, undefined polymers and asphalts, and molten glasses. One reason that pure materials are not used as reference materials is that viscosity usually depends "sig- nificantly and indeterminately on the purity of the materi- al." Any pure material would therefore require purification immediately prior to use as a viscosity standard (e.g., water in ASTM method D 2162). Unfortunately, those materi- als that can be sufficiently purified without elaborate equipment have viscosities of the same order of magnitude as water (Marvin, 1972). Marsh states that "it is necessary that the (calibra- tion) materials show Newtonian behavior, that they are non- -corrosive, and except for molten glasses, have a high solubility in at least two readily available organic sol- vents so as to enable appropriate cleaning." Further re- quirements are stability of viscosity during storage and use, reasonable cost, and safety. The suppliers surveyed included the national standards laboratories of Australia, France, the Federal Republic of Germany, Hungary, Japan, the Netherlands, Poland, the United Kingdom, the United States, and the U.S.S.R. One private laboratory (Cannon Instruments, U.S.) was also included. 26 The viscosities of CRMs available from the 10 suppliers ranged from IOF’to 10n-Pa 3, though the most common range was from 10W3to 30 Pa 3. The uncertainty of the values assigned to standards in the 0.1 to 10 Pa 3 range, and above 10 Pa 3, is about 0.3 percent and 1 percent, respectively. Temperature control necessary to use the standards is generally within 0.01°C for the lower range, and 0.02°C for those standards above 10 Pa 3. 3.5.4.2. Reference Materials in the United States The U.S. NBS supplies three reference materials with viscosities ranging from 10 to IOU-Pa 3. These materials are designed for calibrating rotating cylinder instruments and fiber elongation equipment, and are certified in the temperature range of 464°C to 1545°C. The other major source of certified viscometer calibra- tion materials in the U.S. is provided by the Cannon Instru- ment Company (State College, PA) under contract with ASTM Committee D-2. The materials supplied by Cannon ranges from 0.003 Pa 3 to 900 Pa 3. A discussion of the composition and properties of the Cannon materials is instructive in discerning important properties of calibration fluids. The 0.11 to 1.4 Pa 3 materials, highly refined petroleum oils without additives, are of mixed composition, include naphthinic molecules, and are highly paraffinic. The materials have a narrow, but unspecified, molecular weight distribution. 27 Cannon’s 4.9 Pa 3 to 72 Pa 3 materials are polybutenes or polyisobutenes. Butene is polymerized to the degree necessary to give the appropriate viscosity; they are not blended and do not contain additives. While many polymers are non-Newtonian, polybutenes below about 17000 daltons have been shown to be Newtonian (Porter and Johnson, 1960). The Cannon fluids are probably below 5000 daltons, but may have some slight deviations in Newtonian behavior in the generation of stress normal to that of shear. These devia- tions, however, are too slight to be demonstrated with respect to shear rate (Hardy, 1962). Cannon’s materials have a very temperature-dependent viscosity: a 1°C change in temperature will result in a viscosity change ranging from two to nine percent, depending on the specific oil. The stability of these materials is such that viscosity changes less than 10% per year. While most standards are good for three to five years, Cannon claims a one year shelf-life. The stability estimates are for unopened jars only; an opened jar of oil can pick up solvent vapors and change viscosity (Manning, 1987). A third source, Brookfield Laboratories (Stoughton, MA), supplies silicone oils "traceable to NBS." The oils are purchased in bulk from General Electric Corporation (Water- ford, NY), calibrated, and repackaged. However, Marvin (1972) stated that: "Other series [besides the ASTM-Cannon series]...are available...but none are measured with the degree of precision and control of the ASTM series." Two 28 undesirable properties of silicone oils are that they are difficult to remove from viscometers (Marvin, 1972, and Hardy, 1962), and that their viscosities are "widely known to be slightly shear-thinning" (Manning, 1987). However, Walker (1989) claims that silicone is less likely to become contaminated than petroleum oils, that "chlorothane" effec- tively cleans these fluids from instruments, and that the real shelf-life is 5-10 years for some of the Brookfield silicones. One must note that the Brookfield standards are not part of the national standards organizations’ programs in the U.S. 3.5.5. Rhaameter Manufacturers’ Recommendationa Several instrument suppliers have recommended different practices for assuring the user of the accuracy of measure- ments made with their rheometers. The manufacturers’ advice falls into two categories, described below. 3.5.5.1. Hanging Weights Rotary rheometers constitute a very popular class of instrument sold for viscosity measurement. The instruments’ basic principle of operation is as follows. The sample is held between two independent parts of a sample cell. The most popular geometries are: 1) cone (or plate) and plate, and 2) coaxial cylinders (also referred to as "concentric cylinders,", consisting of a "cup" and a "bob"). One part of the cell is held stationary while the other part is rotated. The resistance of the fluid, held between the two cell components, to this rotation is measured as torque 29 versus the angular velocity of rotation. The viscosity is then calculated from a theoretical understanding of the flow field created in the sample by the sample cell. Either the speed or the torque may be controlled in a rotary rheometer and the other parameter measured. Both the angular velocity and the torque must be accurately known for the resulting viscosity to be accurately calculated. The manufacturers of at least two rotary rheometers (Rheometrics and Haake) recommend that the torque sensor in their instruments be calibrated using "hanging weights." The hanging weight method employs a weight connected to the torque sensor via a thin nylon line which is draped over a suitable disk or cylinder of known radius. The resulting torque can be accurately calculated and related to the instrument torque response. A series of weights and result- ing responses are used to construct a calibration graph. A linear regression of the graph can then be used to translate the instrument readings taken during sample measurement into torque. 3.5.5.2. Calibration Fluids Manufacturers of both relative and absolute viscometers recommend that their instruments be checked occasionally using a standard fluid. Brookfield Laboratories (Stoughton, MA) used a series of calibration fluids to develop the calculation factors for their popular RV series of relative viscometers. Fisons Instruments (Saddle Brook, NJ) sells German-made calibration fluids to those users of Haake 30 viscometers who wish to check the accuracy of their instru- ments. Mitech, Inc. (Willoughby, OH), distributors of the Carrimed Controlled Stress Rheometer (Carrimed Ltd., Dor- king, England), have used Brookfield calibration oils to check the accuracy of the instrument during installation, particularly with regard to the accurate setting of the gap when using the cone and plate geometry. This approach is confounded by the difficulty of temperature calibration at the plate surface for this instrument. Accurate knowledge (within 1% of the true value) of the cone angle and gap between the cone and plate for the Carrimed is exceedingly difficult to obtain, however. 3.6. Alternative Secondary Reference Calibration Materials 3.6.1. Relationship of Fluid Structure to Rheological Prop- erties There has been considerable interest in relating vis— cosity of liquids to their molecular structure. A volumi- nous review of the general literature was made by Bondi (1967). Associations of structure with equilibrium proper- ties (including density, heat capacity, temperature, and pressure) and transport phenomena (relaxation of dipoles) of a wide variety of fluids were discussed, and relationships were derived from fundamental considerations of molecular transport by using "semiempirical correlation schemes" with temperature and pressure. Such a theory could possibly 31 predict ideal calibration materials, in terms of the desir- able Newtonian and handling properties mentioned above. Mathlouthi and Kasprzyk (1984) reviewed theoretical and empirical relationships of sugar-water solution concentra- tion, temperature, and viscosity. Despite this being a relatively simple system compared with many fluids of prac~ tical interest, the authors wrote: "Viscosity of sugar solutions is not easy to put into equations if we take into account the fact that the sucrose molecule is capable of hydration, sugar-sugar association or the promotion of II ‘water-structure’. Forty-six equations relating viscosity to various parameters were discussed. Most were derived from polymer science studies. Most empirical relations used in the sugar industry were found to be adaptations of theo- retical predictions. Any predictions from theory must be confirmed. It seems apparent, therefore, that empirical observations should figure prominently in any scheme to supplement current oil-based calibration fluids with materials more suitable for large-volume applications. For all gases and most low molecular weight homogenous fluids, the shear stress is a linear function of the shear rate, passing through the origin at 7 = 0 (Boger et al., 1985). This class of materi- als conforms to the Newtonian model very well. Solutions and melts of flexible macromolecules, as well as many suspensions, typically decrease in apparent viscosi- ty with increasing rate of shear. The decrease is usually 32 ascribed to the breakdown of a structure at the collodial or molecular level. An example of this change in structure is the behavior of macromolecules, which can become more align- ed and therefore less entangled and resistant to shear with increasing rate of shear. 3.6.2. Survey of Literature Materiala Few fluids have been cited in the literature as having been actually used or proposed as calibration materials for Viscometry. Besides the water standard, petroleum oils, butene polymers, and silicone already discussed, only two other groups of fluids have been seriously considered as Newtonian calibration fluids. The groups can be classified as pure compounds and solutions. 3.6.2.1. Pure Compounda Hardy (1962) noted that the use of pure chemicals as reference standards had been proposed in the scientific community. The rationale held that suitable pure compounds could be characterized by a standards laboratory, then confidently used in other laboratories after appropriate routine purification. Unfortunately, according to Hardy, the techniques for both purification and checking purity of many potential calibration fluids were too specialized for most laboratories. By 1984, however, this situation had changed. Various organic solvents became available with remarkable purities (part per billion levels in some cases) to serve the bur- geoning application of spectroscopy and chromatography 33 analyses. Bauer and Meerlender (1984) studied low viscosity fluids (less than that of water) with the aim of improving the reproducibility (several percent) of measurements of the same fluid on different viscometers. They recommended that several commercially available hydrocarbons and halogenated hydrocarbons be used as viscosity standards. The fluids were commonly available with purity sufficient to charac- terize viscosity to within 0.4%. Few pure compounds that exist as fluids at 25°C have viscosities above 0.1 Pa 3 (Weast, 1971). Glycerin’s vis- cosity is 0.954 Pa 3 but is hygroscopic (Stecher, 1960) and therefore unstable for critical uses. Glycerin was, howev- er, used as one of a series of Newtonian fluids in a study of heat transfer in a food canning process (Anantheswaran and Rao, 1985). Glycerin is available in the U.S. in 570 pound drums for about $1 per pound at a purity of 99.7% (Simpson, 1989). Approximate bulk costs of other fluids per pound in the U.S. are: polybutenes, $2 (Collins, 1987); silicone oils, $8 to $25 (Pucinich, 1989); 67% sucrose syrup, $0.46; corn syrups, $0.11 to $0.20 (Simpson, 1989). Purification and analysis of purity is a critical factor limiting the use of high-viscosity pure compounds as cali- bration materials for Viscometry. Another problem with use of pure compounds is the need to have several fluids of various viscosities available, since many applications require that a range of conditions be evaluated. Ease of use, cost, and safety are also important considerations con- 34 tributing to the disuse of this class of materials as cali- bration fluids for Viscometry. 3.6.2.2. Solutions A true solution is a homogeneous molecular dispersion of two or more substances. In a solution, the individual components are subdivided into particles of molecular size, and can only be separated by physical processes. A solution contains two phases, in which one or more solute(s) are dis- persed in a liquid solvent of some kind, where the solvent is that component which retains its original physical state after the solute(s) are added (Quagliano and Vallarino, 1969). While most viscosity reference materials are true solutions, they consist of two or more fluid components. For purposes of this discussion, let us consider solutions to be solid solutes dissolved in liquid solvents. 3.6.2.2.1. Sucrose The only solution proposed as a calibration material is sucrose-water. Andrade (1947) stated that "The capillary tube viscometer is extensively used for ordinary laboratory measurements of viscosity, being calibrated by the use of a standard liquid such as water or sucrose solution." Bates (1942) reviewed the use of sucrose solutions to calibrate viscometers in the sugar industry. Sucrose solutions were recommended because of a perceived need to reduce errors when using relative viscometers by "employing for calibra- tion purposes liquids with physical properties not very different from the liquids to be investigated." Subsequent 35 discussion by Bates leads one to surmise there were no other calibration fluids besides water available at that time. In fact, NBS began supplying hydrocarbon oils for viscometer calibration in 1938 (Hardy, 1962). Bates (1942) discussed several problems with using su- crose solutions. Sucrose purity was a potential factor, but NBS did (or now does) furnish crystalline sucrose standards. Preparation of the solution was recommended one the same day as use, but filtering (to eliminate dust) of the prepared solution could change concentration. Use of the solution required special care to avoid condensation and evaporation, including a recommendation to work in a temperature range of 15°C to 25°C. Table 1 shows the dependency of solution viscosity on sucrose concentration (Bates, 1942). At moderate concen- trations, the viscosity of sucrose solutions are well under 0.1 Pa 3. Only the most concetrated solutions have viscosi- ties of about 1 Pa 3. Another aspect of the data shown in Table 1 is the effect of sucrose solution concentration on viscosity. The table shows that a slight inaccuracy in the knowledge of sucrose concentration at point of use can lead to a large error in the assumed viscosity of the calibration material. For example, the viscosity of a 73% sucrose solution is only 72% of the viscosity of a 74% sucrose solution. Most "liq- uid sugar" syrups sold to the food industry in the U.S. are 67% sucrose (Hoynak and Bollenback, 1966). Thus it 36 Table-1J.Selected sucrose-water solution. viscosities at 20°C. % Sucrose n (Pa 3) 60 0.0589 61 0.0697 62 0.0830 63 0.0998 64 0.1210 65 0.1482 70 0.4850 71 0.6408 72 0.8609 73 1.178 74 1.643 75 2.344 37 seems reasonable that preparing and using sucrose solutions of viscosity above 0.4 Pa 5 involves considerable inconve- nience and potential error. 3.6.2.2.2. Polymers An almost infinite variety of polymer materials exists. Pure polymers that exist in the fluid state are, with appar- ently few exceptions, soluble only in organic solvents. They are therefore unsuitable for large-volume viscometer calibration uses in the food industry, whatever merits their other characteristics may have. One class of polymers attractive to rheologists are those at least partially soluble in water. Some examples commonly used in the food industry include many polysac- charides such as xanthan, modified celluloses and starches, pectin, various gums, alginates, and gelatin. These materi- als are commonly used to form gels or provide texture to a variety of liquid and semi-liquid products (Sanderson, 1981). Most of these water-associative polymers form solutions that are shear-thinning, with a few being shear-thickening (Szczesniak and Farkas, 1962). Simply stated, these poly- mers are non-Newtonian for one or both of the following reasons. First, they are very large and entangled in solu- tion. The molecular network aligns and deforms in the direction of shear, and this effect increases with rising shear rates. Second, individual molecules of many of these polymers associate with other molecules in solution via 38 hydrogen bonds, ionic attractions, or vanderWaals forces. Some of these associative forces are weak enough to be overcome by shear, even at low rates. Consider the specific example of the widely used gum from the microorganism xanthomonas campestris (Bewersdorff and Singh, 1988). Xanthan gum may be considered to be an anionic polyelectrolyte repeating polymer of pentasaccharide units. The primary structure in solution is formed by the charged trisaccharide side chain folded around a cellulose backbone, yielding a rod-like structure stabilized by intra- molecular non-covalent bonding. The secondary structure undergoes a helix-coil transition depending on conditions of salinity and temperature. In solutions of low ionic strength, the xanthan molecule is a highly extended chain resulting from intra-chain electrostatic repulsions. Adding small amounts of electrolytes to the xanthan gum solution causes the gum to contract and assume a configuration of rigid, worm-like coils, which associate with each other at higher ionic strengths. One way to obtain Newtonian behavior from water-soluble polymers would be to choose those polymers that exhibit little of the above behavior. Specifically, a polymer candidate for a Newtonian fluid would be relatively small, and not form secondary associations with its other polymer neighbors in solution. Many corn syrups are solutions of such polymers, along with smaller saccharides. 39 3.6.2.2.3. Corn Syrupg Corn syrups, known in the United Kingdom as "glucose syrups," have been defined (U.S. Govt., 1977) as: a clarified, concentrated, aqueous solution of the products obtained by the partial hydrolysis of corn starch. Depending on the degree of hydrolysis of the starch, glucose syrups may contain dextrose, maltose, or higher oligosaccharides. The corn industry places a lower limit of 20 DE ("dextrose 0 equivalent,’ a measure of chemical conversion of starch to simpler sugars, using a scale from 0 (no conversion) to 100 (complete conversion) on its definition of corn syrups. This limit differentiates corn syrup from lower conversion products typically referred to as maltodextrins (Junk and Pancoast, 1973). Corn syrups of very high or complete conversion are available that are virtually pure solutions of monosac- charides. Theoretically, these should behave identically to sucrose with respect to shear independency. As the degree of conversion decreases, syrups have increasingly higher concentrations of increasingly larger polysaccharides, thus giving increasing viscosity. In fact, the variety of corn syrups available on the world market is quite remarkable. Table 2 illustrates the range available from one manufactur- er’s line of corn syrups, and some of their properties (A.E. Staley Mfg. Co., 1987). One concern expressed above is the potential of a cali- bration material to gain or lose moisture during use. Corn syrups do have a tendency to take up or lose moisture to the Table 2. DATA 40 Composition and physical properties of A.E. Staley Co. corn syrups. SYRUP TYPE: Staley number 200 300 1300 7300 7350 2300 4300 4400 5500 1CD Dg. of Chiver’n v.1m 1m reg. IQ. mad. uni. hign high “my. v.high DE, % 26 35 43 42 50 54 63 63 -- -- Dextrose, % 5 13 19 9 10 27 37 37 41 50 Fructose, % 0 0 0 0 0 0 0 0 55 42 Maltose, % 8 10 14 34 42 22 29 29 0 3 Maltotriose, % 11 11 13 24 22 15 9 9 0 0 Higher Saochar., % 76 66 54 33 26 36 25 25 4 3-5 Total Solids, % 77.5 80 80.3 80.4 80.7 81 81.6 83.6 77 71 lbisture, % 22.5 20 19.7 19.6 19.3 19 18.4 16.4 33 29 PH 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 3.5 4 41 surrounding atmosphere. This behavior, called hygroscop- icity, is controlled by the syrup’s equilibrium relative humidity (ERH), the relative humidity at which moisture is neither gained or lost. High DE syrups attract moisture because they have a low ERH, while low DE syrups have a high ERH and so will tend to lose moisture to their surroundings (Dziedzic and Kearsley, 1984). Ideally, the viscosity of any material proposed as a calibration fluid should be relatively unaffected by absorption or loss of moisture during calibrating operations. Several workers have made rheological measurements of corn or similiar syrups. Rao and Cooley (1984) used corn syrups diluted with water to various viscosities to deter- mine shear rates in viscometers employing complex geometries such as mixer paddles. The corn syrups were found to be Newtonian when measured with a coaxial cylinder viscometer (Haake Rotovisco). Castell-Perez, et al. (1986) used corn syrups to evaluate a mixer viscometer system, finding the syrups to be Newtonian using a Brookfield coaxial cylinder viscometer. Prilutski et al. (1983) showed that the behav- ior of the Boger fluid (Boger, 1977/8), a solution of poly- acrylamide in corn syrup, was a logical consequence of dissolving a small amount of a high molecular weight polymer in a highly viscous Newtonian fluid. The fluid has a constant viscosity provided by the syrup, yet is highly elastic, the behavior being a sum of the solvent and polymer contributions. Cutler et al. (1983) used syrups similar to 42 those from corn (maple, rosehip, honey, and "pancake" syrup) as Newtonian fluids to study sensory viscosity perception. A cone-plate viscometer (Rheometrics) was used to character- ize the viscosity of the syrups, finding them to be Newtoni- an. Finally, Steffe et al. (1983) found, in a review of the literature, that corn syrups and most fruit juices are New- tonian. Chapter 4 MATERIALS ASTM viscosity calibration oils were obtained from Cannon Instrument Co, State College, PA. Oils used (nominal viscosity, Pa 3 at 25°C) included 860 (0.1), S600 (1.4), S8000 (21) and 830000 (77). Corn syrup samples were donated by A.E. Staley Co., Decatur, IL. Samples (nominal viscosity, Pa 3 at 25 °C) included Isosweet 100 (0.17), Isosweet 5500 (0.80), Staley 4300 (17), Staley 3260 (20), Neto 7350 (47), Sweetose 4400 (87), Staley 1300 (87), and Staley 300 (170). The nominal chemical composition and physical properties for these fluids are listed in Table 2. All samples were used as received. 43 Chapter 5 METHODS With the literature suggesting that corn syrups are Newtonian, the work focused on evaluating the Newtonian behavior of syrups of a wide range of viscosities. Addi- tionally, repetitive analyses were made on the same sample of syrup over a period of days to determine if the syrups changed viscosity as a result of normal laboratory handling. 5.1. Tempepature Considerations The target temperature for all analyses was 25°C. The Carrimed 08500 rheometer’s temperature control was i 0.1°C. The Haake RV-12 rheometer’s temperature control was diffi- cult and time-consuming to adjust, so that a few analyses were conducted at average temperatures as low as 24.0°C and as high as 25.6'C. Most analyses were conducted between 24.8°C and 25.2°C; the temperature within each analysis varied less than i 0.1'C. 5.2. Determining the Degree of Newtonian Behavior Corn syrups were compared with Cannon oil standards (assumed to be Newtonian) for fundamental behavior indica- tive of Newtonian fluids: linearity of shear stress vs. shear rate, insignificant G’, and insignificant time and shear dependency. 5.2.1. Linearity of Shear Stress vs. Shear Rate in Steady M 5.2.1.1. Haake Coaxial Cylindera 44 45 Analyses were performed with a Haake RV12 rheometer (Fisons Instruments, Saddle Brook, NJ) with M-150 and M-500 measuring heads. A Haake F3-K controlled temperature bath supplied oil to a "temperature vessel" which surrounds only the vertical side of the inserted sample cup. The sample cups, designated MV (inner radius = 0.0210 m) and SV (radius = 0.01155 m), include a hole in the bottom through which a temperature probe was inserted, positioned a few millimeters away from the bottom of the inner rotor during analysis. Rotors (the inner, rotating cylinder, also known as "bob") used included the SVI (0.010147 m radius, 0.061468 m height), the SVII (0.010084 m radius, 0.019329 m height), the MVA (0.018402 m radius, 0.05936 m height), and the MVB (0.020053 m radius, 0.061341 m height). The bottom and top surfaces of these rotors were recessed at least 0.015 m. The MVA and MVB rotors were fabricated at the Kellogg Compa- ny and were similar to the MVI and MVII rotors sold by Fisons Instruments. The rotational velocity of the inner rotor during an analysis was programmed using the Haake PG140 controller. Two different types of rotational velocity programs were used: continuous and step-ramp. The first approach (recom- mended by Haake) consists of continuously increasing the rotational velocity of the bob from the initial resting state to the maximum velocity desired. The step-ramp method quickly increases ("steps") rotational velocity to a desired value, then holds it there for a length of time to establish 46 equilibrium. The velocity is then stepped to the next desired value, and held again. The various steps span the desired shear rate range. The choice of sample fixtures (rotor and cup) and rota— tional velocities was dictated by the maximum torque gener- ated by turning the rotor against the various viscosity fluids tested. The maximum torque capability of the measur- ing heads limited the shear rates available to under 500 s'"1 for the lowest viscosity materials. Table 3 shows the rotors used, their dimensions, and the ratio, R, of the rotor to cup radii. The Haake RV12 was connected to an IBM-PC for data capture and subsequent analysis. A Data Translations, Inc. (Marlborough, MA) DT2801 board installed in the IBM-PC converted the rheometer’s analog signals of rotational velocity, torque, and temperature to digital form, and stored them to a disk at 2 second intervals. Viscosity results were calculated from the digitized data using Lotus 123 (Lotus Development Corp, Cambridge, MA). The Lotus worksheet calculated shear rate from rota- tional velocity using a generalized method that does not assume a fluid model a priori (Krieger, 1968). Step-ramp 47 Table 3. Haake concentric cylinder dimensions. Rotor Radius (m) Height (m) R MVA 0.01840 0.05936 0.876 MVB 0.02005 0.06134 0.955 SVl 0.01015 0.06147 0.879 SV2 0.01008 0.01933 0.873 48 experiments used step intervals of 30 seconds duration, and were calculated using the average of the last 16 seconds of each step. Continuous-ramp experiments used all points of data. 5.2.1.2. Carrimed Cone and Plata The Carrimed Controlled Stress Rheometer CS500 (Carri- med, Ltd., Dorking, England) was used with cone and plate geometry. Various cone angles and diameters were used: 1°- 4cm, 1°-2cm, 2°-2cm, and 2°-6cm. The Carrimed was supplied ' a device designed to elimi- with an optional "solvent trap,’ nate sample drying at the edge of the cone-plate gap. Figure 3 shows a diagram of the cone and plate with the solvent trap in place. An appropriate solvent (water was used for corn syrup samples) was placed in the well sur- rounding the cone’s shaft, and the solvent trap was lowered in place. The solvent formed a barrier to air circulation between the air gap inside the trap and outside. This solvent trap was not used during any experiments with the Cannon oils, but was used for most corn syrup experiments. The Carrimed was programmed to increase torque in a continuous program (analogous to that of the Haake instru- ment) and measure the resulting rotational velocity. Achieving a fixed maximum shear rate was therefore a trial and error process. All replicate experiments used identical torque programs. 49 CV/ SOLVENT TRAP Fluid Flui Figure 3. Carrimed cone and plate geometry with solvent trap. 50 TX). The software both controlled the Carrimed instrument and calculated experimental data. Shear rate and shear stress were calculated for each of the approximately 200 data points available for a typical experiment of two min- utes duration. A straight line with non-zero intercept was fit to the data using the method of least-squares, and a correlation coefficient, r“, was calculated. The correla- tion coefficient and magnitude of the non-zero intercept were used as indicators of the degree of Newtonian behavior. 5.2.2. Dygamic Oscillatory Shear Experiments The Carrimed C8500 was also used for dynamic oscillatory tests. The reason for using this testing mode was to check for elastic behavior. Fluids are known to exist that are very elastic, yet whose steady shear viscosity does not change with shear rate (Boger, 1977/78). The Carrimed instrument had an oscillation frequency range of 0.01 to 10 Hz, while its torque range stretched to 50 N m. Two types of experiments were conducted: torque sweeps at constant frequency, and frequency sweeps at con- stant torque. The amplitude of oscillation was automati- cally limited to 10 milliradians via software control. Tor- que sweeps used a frequency of 0.5 Hz (in the logarithmic middle of the available range), while the frequency sweeps were conducted from 0.05 to 4.332 Hz, the widest efficient range available on the CS500. 51 5.2.3. Time-Dependency A truly Newtonian fluid not only has a viscosity which is constant with varying shear rate, but also with repeated sampling over time (non-time-dependent). The tendency of test fluids to change with continuing or sequential analysis was therefore studied. Additionally, an ideal calibration material should have a long shelf-life, i.e., should main- tain constant properties for useful periods after manufac- ture. Viscometric analyses of the same fluid were therefore performed on different days. 5.2.3.1. Multiple Analysea of Same Sample Load Most steady shear analyses were performed with an up- -ramp (increasing rate of shear) immediately followed by a down-ramp. A single shear stress - shear rate plot of both ramps should show identical results with no thixotropic loop present. A second approach taken was to simply perform a series of identical analyses on a sample without removal from the rheometer. A few analyses were conducted after leaving the sample in the rheometer for a period of one to several days, allowing for the possibility of structural recovery in the sample. 5.2.3.2. Separate Analyses Conducted Days or Months Apart on the Same Sample Samples were repeatedly analyzed over a period of months to determine shelf stability. Noting that the current ASTM calibration oils have a shelf-life of one year in a sealed 52 jar, comparisons were made between the differences among successive oil analysis results, and the corresponding differences differences among corn syrup analyses. Chapter 6 RESULTS AND DISCUSSION 6.1. HaakaaSensor Windmp The RV-12 measuring head uses a spring to measure the torque exerted on the fluid by the cylinder rotating against the fluid being tested. The measuring spring requires an angular displacement to register torque. In an experiment covering a range of rotational speeds, the torque-measuring spring will deflect to differing degrees, resulting in the "windup effect" (Whorlow, 1980; Steffe, 1987). The windup effect is shown in Figure 4, a plot of torque vs. rotational speed for two typical continuous-ramp Haake experiments. Each experiment was conducted by increasing the sensor’s rotational speed at a constant rate of increase from an initial resting state to the maximum speed, then back down again. These two parts of a single experiment are called the "up-ramp" and the "down-ramp." A perfectly Newtonian fluid, measured using a perfectly stiff torque sensing device, should show identical traces for both the up and down portions of each experiment. The graph shows, however, that the down-ramp trace is above that of the up-ramp. A time-dependent fluid, one which would be expected to break down over the course of this experiment, would be expected to give a down-ramp below that of the up-ramp. It was established above (Section 3.5.4.2.), however, that the Cannon oils are Newtonian, 53 54 100 down—ramp 50 i /'1 ’3 Cannon 830000 oil / '3 1t down-ramp O / m 60 - / "a I / H 1:: 4L . / g 40 + //up—-ramp / Cannon 3600 oil 5° / ..a ‘r / / / 20 i A up-ramp / . 0 i r % Wu l L i . 0 50 100 150 200 250 rotational speed (rpm) Figure 4. Haake "windup" effect. 55 certainly true for this experiment conducted at shear rates below 300 s-L The anomalous behavior shown in Figure 4 may be explain- ed by either (or both) of two mechanisms. First, consider the windup effect mentioned above. The sensor, being con- stantly rotated faster on the up-ramp, never comes to equi- librium until the point when the rotational speed begins its descent during the down-ramp portion. Then, the torque sensing spring is always wound too tightly during the down- -ramp. The result is that the up-ramp results are too low, while the down-ramp results are too high. A second possible source of the effect of the response lag could be the inertia of the fluid and rotating sensor, but the magnitude of such error should be small. Sample and/or rotor densities or sizes (and thus masses) would need to be varied to experimentally test such a hypothesis, because the two effects cannot otherwise be separated. It seems easier to use the rotor rotation programming technique below, which will compensate for both error sources. The continuous increase of sensor rotational speed is the manufacturer’s recommended method for conducting "abso- lute viscosity" measurements over a range of shear rates. One way to eliminate the windup effect is to take data only once the torque spring has achieved equilibrium by conduct- ing a step-ramp experiment. Use of this technique has been reported (Rao et al., 1975). Figure 5 shows this behavior for the Haake RV-12 sensor during an experiment that 56 4O , - r +— 70 l i 1 4- 60 30.L +l % T 50 E E " t 3 a ‘g L #40 g Q 20 d I 5 CL 0. U“ 1’ 30 8 1 o: r l .5. 0 +20 m 10 '1— } I9 f 4F 10 O 1* ,L f , + l + f, 4. j‘ r 0 0 100 200 300 400 500 600 ANALYSIS TIME (s) Figure 5. Haake step-ramp program. and sensor lag, Cannon $30000 oil. 57 included both an up-ramp and a subsequent down-ramp analysis of Cannon 830000 oil using the SVII geometry. The lag of torque response behind the rpm program is small but does illustrate the error due to the twisting of the measuring spring. 6.2. Degree of Newtonian Behavior 6.2.1. Steady Shaam 6.2.1.1. Haake Coaxial Cylinders One measure of the Newtonian behavior of a fluid is the linearity of a shear stress - shear rate plot. Table 4 com— pares a Cannon calibration oil, S60, with a Staley corn syrup, Isosweet 100. Analyses were conducted on the Haake using the MVB coaxial cylinders, continuous or step-ramps to a maximum rpm of 200 or 250 for the oil or syrup, respec- tively. Only the up—ramp (increasing rotational velocity) data are included. Syrup analyses employed a cover for the top of the sample holder to reduce drying during the analy- ses. The table shows that the Staley corn syrups have corre- lation coefficients approximately equal to that of the accepted Newtonian standards. Another way way of examining the same analytical re- sults is via a plot of apparent viscosity na_vs. V, where “a = o/y. A perfectly Newtonian material, in an ideal viscome- ter, should yield a na'vs. y plot that is a straight line horizontal to the x axis. Figure 6 illustrates this type of plot comparing single analyses of syrup and oil 58 Table 4. Linearity of shear stress - shear rate analysis results as expressed by correlation coefficient, r2, for low viscosity fluids, concentric cylinders. NEWTONIAN BINGHAM MATERIAL sm, 3'1 r2 «1, Pa 3 cy, Pa r2 Continuous ramp data: C. 860 592 0.999 0.111 -0.892 0.999 " 596 0.999 0.0996 -0.584 0.999 " 596 0.999 0.0974 -0.688 0.999 I.S. 100 478 0.999 0.181 -0.611 0.999 " 478 0.999 0.180 -0.283 0.999 Step ramp data: C. 860 596 0.999 0.0980 -0.288 0.999 I.S. 100 478 0.999 0.179 -0.029 0.999 59 taken from Table 4. The corn syrup behaves very similarly to the oil. Both fluids show an abrupt increase in viscosi- ty at low shear rates due to the windup effect. Because the spring lags behind the rotation program of the inner rotor, the reported rotational velocity is greater than it actually is. The fluid therefore "appears" to be less viscous than it really is during this early portion of the experiment. Another manifestation of this effect is that the oppo- site shape of the curve occurs in down—ramp experiments (not shown). Both fluids show relatively constant viscosity at the higher shear rates, as expected for a Newtonian fluid. Similar comparisons for medium and high viscosity fluids follow. Table 5 has both medium and high viscosity data included. Analyses were performed using the SVII coaxial cylinder system, with step-ramps of 50 to 100 max- imum rpm. The corn syrup data is as linear as that of the calibration oils. 6.2.1.2. Carrimed Cone and Plata The cone and plate geometry has several advantages over coaxial cylinders. Implemented in the Carrimed CSSOO con- trolled-stress rheometer, the geometry is very easy to load. One simply places a measured amount of liquid on the plate surface, then raises the plate to a close, pre-determined proximity with the cone. An excess of fluid, easily visual- ly inspected, is carefully wiped or scraped away, leaving an optimal sample load between the moving cone and the fixed plate. 60 0.200 STALEY I.S. 100 CORN SYRUP APPARENT VISCOSITY (Pa 8) O 8 O l 1 .° CANNON 860 CALIBRATION on. // I 005011; I 11 0.000 . 4 t + . 4 , if , l , 0 100 200 300 400 500 600 l SHEAR RATE (sec— ) Figure 6. Linearity of low viscosity fluids, concentric cylinders. 61 Table 5. Linearity of shear stress - shear rate analysis results as expressed by correlation coefficient, r“, for medium and high viscosity fluids, concen- tric cylinders. NEWTONIAN BINGHAM MATERIAL fl, Pa 3 r2 n, Pa 3 (5" Pa r2 C 8600 1.33 0.999 1.36 -1.77 0.999 S. I85500 0.829 0.999 0.840 -1.39 0.999 " 0.812 0.999 0.836 -2.59 0.999 C. 830000 73.8 0.997 72.1 43.3 0.999 " 70.2 0.998 68.8 37.6 0.998 " 76.8 0.999 77.1 -6.38 0.999 S N7350 51.9 0.997 50.0 45.8 0.999 " 47.2 0.997 45.2 46.6 0.999 " 53.3 0.998 52.0 31.9 0.999 S. S4400 87.0 0.999 85.8 19.7 0.999 S. 1300 82.7 0.999 81.4 21.0 0.999 " 87.2 0.999 86.1 17.3 0.999 62 Another major advantage is that, for any given angular velocity of the spinning cone, the shear rate, and conse- quently the shear stress, experienced by the fluid in the gap is the same throughout the gap (Whorlow, 1980) as long as the cone angle 0 is less than 5°. The calculation of shear rate is then simply y = o / 0 and the construction of the shear stress-shear rate curve is straightforward and linear across a wide range of shear rates. The uniform shear rate advantage, together with the CSSOO’s constant stress operating-measuring principle (as opposed to the Haake’s torsion spring), should result in well-behaved flow curves. Figure 7 shows a “a vs. 7 curve for relatively low viscosity fluids. The Staley IsoSweet 100 corn syrup’s flow curve has a shape very similar to that of the somewhat more viscous Cannon $600 calibration oil. Similarly, Figure 8 compares flow curves for high viscosity fluids. The curves, again, have similar shapes, suggesting that the corn syrups are as Newtonian as the accepted ASTM oils when measured in this way. The large amount of scatter seen in this figure is probably due to a slightly warped or misaligned cone and plate fixture. The high initial results seen in both figures is most likely due to fluid and instrument inertia, as no torque measuring spring is used in the C8500. A potential lag in response from applied torque and measured angular displacement, 63 1.6 4 J» ‘ CANNON $500 OIL U) :3 t E i 8 1.0 3r O S *2 l ‘62 0 7 + < . 0. (1. it “ l 0.4 «l- STALEY I.S. 100 SYRUP 0.1 h 4. l A. % + 4 : 4 .L 4 .L 4 : 1r r o 10 20 30 40 50 60 7o 80 SHEAR RATE (3“) Flow curves of lower viscosity fluids, plate. Figure 7. cone and 64 200 l 19.5 4- 3? J} G &. 19.0 + E CANNON S600 01L 8 .. 0 VJ 5’ 18.5 d)- i l m L % 18.0 1 it STALEY 3260 SYRUP 17.5 1r- 4» 17.0 . % A ,4 4 .L 4 4 + 4 .L : ; ¢ 4 0 50 100 150 SHEAR RATE (s‘l) Figure 8. Flow curves of higher viscosity fluids, cone and plate. 65 caused by the finite digital sampling frequency of the data collection electronics, should be negligible. Finally, consider Table 6, a comparison of linearity of flow curves of oils and syrups generated in the Carrimed’s cone and plate geometry. The correlation coefficients, r2, for syrups compare very favorably with those of the oils. Once again, the syrups show evidence of Newtonian behavior similar to that of the currently accepted calibration mate- rials. 6.2.2. Dynamic Oscillation As discussed in Section 2.1.2., dynamic oscillatory experiments can be used to check for a fluid’s solid-like behavior in a way that steady shear analyses cannot. The dynamic viscosity, fl’, should maintain the same value across a range of frequencies, and have the same value as the steady shear viscosity. At the same time, the value of tan (6) should be much greater than 1, since the fluid prop- erties should predominate over the solid properties. Vary- ing oscillation frequency is an experimental analog to vary- ing shear rate in steady shear. Finally, n’ should have the same value as n (steady shear viscosity). The Carrimed oscillation results, while somewhat errat— ic, show the behavior expected of a pure liquid. An example is shown in Table 7, which contains the results of simple 0.05 to 4.332 Hz frequency sweeps of Cannon $8000 oil using an amplitude of 0.01 radians. One can see that the n’ re— sults were reasonably consistent, and compared reasonably to 66 Table 6. Linearity of shear stress - shear rate analysis results as expressed by correlation coeffecient, r2, cone and plate. MATERIAL n, Pa s r2 Staley I.S. 100 0.195 0.999 " 0.161 0.999 Cannon S600 1.25 0.999 " 1.12 0.999 " 1.29 0.999 " 1.29 0.999 Staley S. 4300 17.3 0.999 Staley 3260 19.9 0.999 " 19.4 0.999 " 17.4 0.999 Cannon S8000 18.7 0.999 " 17.6 0.999 Cannon 830000 77.8 0.999 " 78.3 0.999 " 74.9 0.999 Staley 4400 101 0.999 67 the fluid’s calibration certificate: an average of the results was 8.9% higher than expected. At the same time, tan (6) values were very high, being somewhat erratic. Table 8 is a comparison of Carrimed oscillation results of standard oil and a syrup of similar viscosity. Note that while the data is again somewhat erratic (probably due to the CS500’s limited computer digital feedback control rate, significant instrument and fluid inertia, mechanical reso- nances, and slight sensor misalignment) the syrup did show increasing G’ values with increasing frequency, whereas the oil did not exhibit this behavior. One may conclude that while the dynamic viscosity was constant over the range studied, the syrup had a significant elastic component. One possible cause of this behavior was the drying of the syrup at the edge of the cone and plate sample cell. Drying could occur despite use of the solvent trap accessory, since a brief interval was inescapably required to position the solvent trap after loading the syrup between the cone and plate. The dried syrup would show significant elasticity; no analogous structure would form when testing an oil. 6.3. Time and Shear-Dependency 6.3.1. Up-Ramps vs. Down-Rampa All Carrimed steady shear analyses and most Haake analyses were performed by increasing rotational velocity from zero to a maximum, then decreasing rotational velocity from the maximum to zero. These two parts of an analysis should yield identical results if the fluid does not change 68 Table 7. Carrimed dynamic oscillation results for Cannon S8000 oil at a constant amplitude of 0.01 radians. Frequency, Hz tan (6), rad/s n’, Pa 3 0.0500 -229 22.9 0.0661 -79 23.8 0.0873 2480 23.3 0.115 110 23.1 0.153 720 23.4 0.202 51 23.3 0.264 170 23.1 0.352 61 22.8 0.465 151 22.9 0.615 124 22.7 0.813 132 22.9 1.07 4480 22.6 1.42 207 22.7 1.88 127 22.5 2.48 79 22.5 3.28 173 22.4 4.33 272 22.3 69 Table 8. Carrimed dynamic oscillation results for Cannon S8000 oil and Staley 3260 syrup. Cannon $8000 oil Staley 3260 ayrup Freq (Hz) G’, Pa fl’, Pa s 6’, Pa n’, Pa 3 0.0500 0.106 22.6 0.236 26.0 0.0661 -0.121 23.0 0.281 25.7 0.0873 -0.154 23.0 0.483 25.2 0.115 0.0221 22.7 0.551 24.6 0.153 -0.647 22.5 0.700 24.8 0.202 -0.796 21.9 1.12 24.4 0.266 0.0104 22.3 1.46 24.2 0.352 -0.364 22.3 2.39 24.2 0.465 -0.122 21.9 2.50 23.6 0.615 0.0329 22.1 3.99 23.7 0.813 0.954 22.3 4.52 23.9 1.07 0.0482 21.9 6.65 23.4 1.42 0.664 22.3 10.6 23.2 1.88 -1.12 22.1 14.6 22.9 2.48 -4.63 21.8 16.6 22.6 3.28 -9.25 20.8 30.8 21.4 4.33 -25.4 19.5 45.6 20.6 5.73 5.37 0.0 46.7 20.2 7.57 15.8 0.780 92.5 19.7 10.0 49.1 1.53 40.0 19.7 70 with time or shear history. A thixotropic loop is typically seen in fluids that break down during the first part of the analysis. A hypothetical example of a thixotropic loop for a shear-thinning material is shown in Figure 9. A proper thixotropic loop is completely reversible, whereas a non- reversible decrease in viscosity is called rheomalaxis (Whorlow, 1980). The Carrimed in steady shear mode was used to check for thixotropy. The Haake was not used due to its "wind-up" property which prevented accurate comparison of up- and down-ramps. Typical results for medium and high fluids are shown in two plots, Figure 10 and Figure 11, respectively. No significant thixotropic loop is evident from the graphs; the somewhat jagged curves are probably due to a slight warp or misalignment in the Carrimed’s fixtures. Results for the low viscosity Staley Isosweet 100 syrup were unusual. Figure 12 compares the syrup with a Cannon 860 standard oil. While the oil shows essentially no dif- ference between the up and down curves, the syrup shows a down loop that first drops below the up loop at the higher shear rates, then crosses so that the down loop is above the up loop at the lower shear rates. This anomalous behavior is probably due to drying of the syrup at the edge of the cone and plate. The lowest viscosity fluids were analyzed without the benefit of a solvent trap. Drying certainly would also have happened with the higher viscosity syrups had a solvent trap not been used during those analyses. 71 UP‘RAMP // DOWN RAMP Y Figure 9. Theoretical thixotropic loop. 72 Shear Sress (Pa) o A l A l 1 l T 1 r I I I 0 50 100 150 Shear Rate (3’ 1) Figure 10. Flow curves of medium viscosity fluids, cone and plate. 73 Stoley Sweetose 4400 syrup Cannon $30000 oil 3000 T l l :3 2000 4 it (D (D 8 i (73 B Q) 5, 10001» 1 l o 0 Figure 11. J . I r ' T 10 20 3O 40 Shear Rate (3’ 1) Flow curves of high viscosity fluids, cone and plate. 74 16 Staley I.S. 100 Shear Stress (Pa) Cannon S60 I I r T 0 20 40 80 80 Shear Rate (rs-1) Figure 12. Flow curves with thixotropic loop of low viscosity fluids, cone and plate. 75 6.3.2. Repeated Measures on Same Sample Load The tendency of a fluid to degrade with time and with shear over time was tested by analyzing the same sample of fluid repeatedly. The sample was simply left in the test geometry between test runs. A comparison of Haake analysis results for several fluids is found in Table 9. The table includes the Newtonian viscosity of each of several test runs performed on several fluids, as well as the percent difference by which subsequent results vary from the first result. All data reported are from the up-ramp part of the analysis only. The data show that the standard calibration oils tended to decrease in viscosity with repeated analysis. The sim- plest explanation for this behavior is slight viscous heat- ing during each analysis, a phenomenon in which some of the energy applied to cause flow was converted to heat, raised the temperature and thereby slightly lowered the viscosity for subsequent analyses. For example, data taken from the Cannon 88000 oil’s bottle label shows that an increase from 25.00°C to 26.00°C would result in a viscosity drop of 7.8%. The Haake’s removable outer coaxial cylinder was enclosed in an aluminum container through which recirculating oil was pumped, but the heat transfer was less than perfect. More- over, the walls of the narrow-diameter SV cup were approxi- mately 10 mm of solid stainless steel, reducing even more the ability of the circulating fluid to maintain constant sample temperature. The Haake rheometer is typical of 76 Table 9. Results of analyses repeated on the same load of various oils and syrups, concentric cylinders. Fluid run no. n (Pa 3) % Difference S60 oil 1 0.0977 2 0.0973 -0.4 3 0.0977 0 I.S. 100 syrup 1 0.175 2 0.177 0.7 3 0.177 1.1 88000 oil 1 20.6 2 19.6 -4.7 3 19.3 -6.3 Neto 7350 syrup 1 51.9 2 51.8 -0.1 3 51.8 -0.1 $300000 oil 1 76.8 2 75.9 -1.2 3 75.9 -1.2 1300 syrup 1 82.7 2 83.7 1.2 3 84.6 2.4 77 many coaxial cylinder instruments, and rather complicated corrections for this effect have been published (Whorlow, 1980). Corrections for cone and plate geometries are even more involved than those for coaxial cylinders because the thickness of the sample fluid varies with radius. Table 9 shows that corn syrups behaved differently than the standard oils. Whereas the oils’ viscosity decreased slightly with repeated analysis, the syrups’ viscosity in- creased slightly, except the Neto 7350 syrup, which was essentially unchanged. The syrups probably increased due to drying at the edge of the coaxial cylinders. This drying occurred despite use of a cover plate, positioned 5-10 mm above the top of rotating cylinder. The cover plate had a hole slightly larger than the rotor’s shaft, and certainly allowed some contact of the sample with circulating ambient air. The increase seen was small, but did point out the sensitivity of corn syrups to drying during analysis. It is also possible that the syrups could absorb water both during transfer to the sample fixture and during testing. Perhaps the edge drying and moisture gain effects canceled each other out for the Neto 7350 syrup. Only a controlled study would differentiate these effects. 6.3.3. Repeated Meaaures on Different Sample Loaaa An important requirement for viscosity calibration fluids is that they remain stable over a reasonable period of time. Several fluids were therefore analyzed on differ- ent days using the Haake with coaxial cylinders. To 78 accomplish this, samples were re-loaded each day. Multiple loads were run on some days to avoid confounding the effect of sample load with that of different days. Multiple runs (analyses) were run for most sample loads. Up-ramp viscosi- ty results, including averages and the results of statisti— cal comparisons, are shown in Table 10. Student’s t test (for comparison of a single pair) or Dunnett’s t test (for multiple comparisons within a single fluid type) were used to compare subsequent day’s analyses with those obtained initially for a given type of fluid. The probability of a Type I error was 0.05 (Gill, 1978). The table shows that the standard oils were generally more different from initial results on subsequent days than were the syrups. Only the Neto 7350 syrup was not signifi- cantly different on a following day. The Staley 1300 syrup smelled slightly fermented when analyzed the second time; all samples were stored at ambient temperature, approximate- ly 22°C. The addition of a suitable preservative, such as potassium sorbate, could extend the shelf-life of the syrup, while not materially changing the syrup’s rheological prop- erties. Certainly, the present study is not a definitive evaluation of the reproducibility of Haake viscosity analy- ses. However, one can see that the syrups, when not spoil- ed, were no less reproducible than the standard oils during the period studied. Table 10. 860 oil I.S. 100 syrup $30000 oil Neto 7350 syrup 1300 syrup 79 Results of analyses repeated on different sample loads of various oils cylinders. Up Ramp 11 (Pa 8) 0.109 0.0977 0.0973 0.0977 0.0952 0.0951 0.0960 0.0959 0.173 0.175 0.177 0.177 0.179 0.178 0.178 0.179 73.8 72.8 72.1 70.2 69.6 69.1 76.8 75.9 75.9 78.1 77.4 51.9 51.8 51.8 47.2 46.4 53.3 53.7 51.5 54.7 82.7 83.7 84.6 53.7 Days Run After Test no. 20 95 138 185 HODNI-‘QOJNHm-BQNHNHQNHmm-waH-BWNI—‘fiQDNH-D-CDNHODNHH 185 and syrups, concentric Avg % Sig. n (Pa 3) Diff. Diff.? 0.0976 -10.0 yes 0.0956 -11.9 yes 0.175 0.177 0.179 00179 -103 yes 72.9 69.7 76.2 6.9 yes 77.7 9.1 yes 52.8 46.8 53.3 0.9 no 83.7 80 6.4. Sample Stability in Handling The Cannon fluids proved to be very stable during the typical manipulations required for laboratory experimenta- tion over a period of months. Since solvent vapors can potentially dilute stored oil (Manning, 1987), care was taken to clean viscometer fixtures and sample handling equipment well away from sample bottles. The most important stability concern when handling corn syrups was drying on exposure to air. A slight skin was noticeable at the fluid’s surface in the sample container when spooning or pipetting most samples, being more conspic- uous for the more viscous fluids. Stirring the sample before removing a specimen for analysis seemed to alleviate any negative effects of the skin in the container itself. Sample drying during analysis is a common problem in Viscometry, as has been alluded to several times in the above discussions with regard to corn syrups. Of the in- struments used in this research, the Carrimed’s cone and plate geometry was by far the most susceptible to sample drying. Figure 13 shows a comparison of steady shear flow analyses performed on the Carrimed for Sweetose 4300 syrup. The loop on the lower right results from an up- and down- -ramp test using the solvent trap attachment for the cone and plate fixture. The syrup shows a slight thixotropic loop, as the down-ramp portion is below that of the up—ramp. The upper loop is the same syrup (different sample load) 81 3000 J— rumour l SOLVENT TRAP ; 2000 + / WITH 6 9: SOLVENT TRAP V) E l’ I; a 1000 I a T/ O 4. # 1F 4 4 ‘f ¢ l 4 4. ‘r % } *fi 0 50 100 150 SHEAR RATE (2.“) Figure 13. Flow curves of Staley Sweetose 4300 corn syrup, effect of solvent trap on cone and plate results. 82 using the same cone, but without the solvent trap. The up-ramp’s curve turns up, almost certainly due to drying during the 30 second analysis period. Confirming this conclusion, the down-ramp’s curve is well above the up-ramp, showing such a large degree of drying that the viscosity (the slope of a line tangent to the curve at any point of interest) continues to increase as the torque is continuous- ly decreased. The large effect of the solvent trap discussed above was opposite in both trend and degree to that expected for a fluid that is not sensitive to air. One would expect a small amount of drag from the solvent held in the solvent trap’s well (see Figure 3). A test of the solvent trap’s potential effect was made by first performing three succes- sive up-down ramp analyses on one sample load of Cannon S8000. The solvent trap was then positioned without dis— turbing the sample between the cone and plate, and water was added to the trap’s well. A second set of three analyses was then conducted. The mean plus or minus one standard deviation for the "without trap" and "with trap" results (six results per trap status) were 22.02 i 0.47 and 21.49 1 0.43, respectively. These means are not significantly different in a t test at p = 0.05. The trend in the data is unexpected, and may be due to slight sample loss or warming of the sample. The stability of corn syrup viscosity over extended periods of corn syrups was not systematically studied. Some 83 of the high fructose syrups require storage above ambient to minimize crystallization and "color development." A.E. Staley Co. recommended that Isosweet 100 be stored at 35°C to 40.5°C. Samples stored at 22°C did form crystals, but these were dissolved by heating at 37°C overnight. The data presented in Table 10 suggest that most syrups should have stable viscosities for several months. Chapter 7 CONCLUSIONS Corn syrups were shown to have potential for use as Newtonian calibration fluids in some applications where the established standard oils are not suitable. Specifically, these applications include food industry pilot and plant calibration operations that require large volumes of test fluids and subsequent water-based cleanup. Examples of potential calibration uses are tube viscometers, pumping systems, process viscometers, mixers, manifolds, filters and screens, heat exchangers, and extruders. The syrups were shown, under the conditions tested, to have viscosities as constant with varying shear rate as those of standard ASTM calibration oils. Some syrups showed significant elasticity that may be due to their tendency to dry upon exposure to air. Appropriate precautions in handl- ing and analysis were able to reduce this effect to insig— nificant levels for many practical food industry purposes. The stability of corn syrup viscosity was apparently several months but probably well under one year. Those wishing to use syrups as calibration fluids should carefully characterize not only the material with respect to its application, but also the application itself. Properties of the fluid that must be determined include steady shear viscosity and temperature sensitivity, elastic- ity, sensitivity to air drying, and shelf-life. The appli- 84 85 cation should be characterized as to needed viscosity and temperature range, fluid exposure to air, and the length of time needed to conduct the calibration. Chapter 8 SUGGESTIONS FOR FURTHER STUDY Several areas of study should be pursued by those wishing to use corn syrups as calibration fluids. First, a systematic study, using control charts, should be made of the shelf-life of the chosen fluid. This study would be made using a laboratory viscometer of reasonably small volume so that a Cannon standard oil of similar viscosity could be included for comparison. Second, a demonstration of use of the syrup in a tube viscometer or other calibra- tion pumping apparatus would: a) provide experience in handling the syrups; b) allow development of techniques for minimizing drying and other deleterious effects; and, c) help understand the specific relationship between the syr- up’s tendency to dry and its rheological behavior in the laboratory and in the pumping apparatus. Finally, an inqui- ry should be made into the effect, if any, of the syrup’s elasticity on the chosen application. The author’s professional experience with developing control samples for rheological analyses suggests the fol- lowing: anyone considering using calibration fluids should have at least some understanding of statistical process con- trol and control sample charting. Personal experience with developing control sample programs for rheological analyses suggests that simple one-point calibration checks of equip- ment can be severely misleading as to the accuracy and 86 87 precision of results obtained with such equipment. In fact, appropriate control charting frequently demands that routine calibrations mat be carried out (Anderson, 1989). Another important use of statistics in calibration work is the design of the experiments and subsequent analysis of results obtained. Careful random sampling will minimize systematic bias, maximize productivity, and thus lead most directly to correct conclusions about the magnitude and sources of variation in the system being studied. Proper control charting, ideally in consultation with an experi- enced statistician, is the only simple way to achieve an accurate estimate of the reliability of a potential calibra- tion fluid and the equipment with which it is used. B I BL IOGRAPHY BIBLIOGRAPHY A.E. Staley Mfg. Co. 1987. Analyses of Staley corn syrups. Technical Data Sheet 110 384011, Sweetener Business Group, A.E. Staley Mfg. Co., Decatur, IL. American Society for Testing Materials. 1973. Standard method of basic calibration of master viscometers and viscosity oil standards. D 2162-64. ASTM Committee D-2, Philadelphia, PA. American Society for Testing Materials. 1987. Standard method of basic calibration of master viscometers and viscosity oil standards. D 2162-79. ASTM Committee D-2, Philadelphia, PA. American Society for Testing Materials. 1983. Standard test method for kinematic viscosity of transparent and opaque liquids (and the calculation of dynamic viscosi- ty). D 445-83. ASTM Committee D-2, Philadelphia, PA. Anantheswaran, R.C., and Rao, M.A. 1985. Heat transfer to model newtonian liquid foods in cans during end—over-end rotation. J. Food Engr. 4:1. Anderson, S. 1989. private communication. Statistician, Kellogg Co., Battle Creek, MI. Andrade, E.N. da C. 1947. Viscosity and plasticity. Cam- bridge University Press, Cambridge, England. Bates, F.J. 1942. Polarimetry, saccharimetry and the sugars. Circ. NBS C440. U.S. Govt. Printing Office, Washington, DC. Bauer, H., and Meerlander, G. 1984. Precise viscosity mea- surements of newtonian liquids with v < 1 mmz/s for the selection of suitable standards. Rheol. Acta 23:514. Bewersdorff, H.W., and Singh, R.P. 1988. Rheological and drag reduction characteristics of xanthan gum solu- tions. Rheol. Acta 27:617. 88 89 Bingham, E.C., and Jackson, R.F. 1919. Standard substances for the calibration of viscometers. Nat. Bur. Std. (U.S.) Bull. 14:59. Boger, D.V. 1977/78. A highly elastic constant-viscosity fluid. J. Non-Newt. Fluid Mech. 3:87. Boger, D.V. , Denn, MOM. , and Binnington’ ROJO 1985. Char- acterization and flow of rheologically-complex liquids. p 2.3. Society of Rheology, Inc., New York Bondi, A. 1967. Viscosity and molecular structure. In: "Rheology: Theory and Applications," F.R. Eirich, ed. Academic Press, New York. p1. Cali, J.P, Mears, T.W., Michaelis, R.E., Reed, W.P, Seward, R.W., Stanley, C.L., Yolken, H.T., and Ku, H.H. 1975. The role of standard reference materials in measurement systems. U.S. Natl. Bur. Stand. Monograph 148. Wash- ington, D.C. Cannon Instrument Company 1988. Catalog and price list. State College, PA. Cannon, M.R. 1944. Viscosity measurement, master viscome- ters. industrial and engineering chemistry. Analytical Edition, IENAA. 16:708. Castell-Perez, M.E., Steffe, J.F., and Morgan, R.G. 1986. Adaptation of a Brookfield (HBTD) viscometer for mixer Viscometry studies. J. Texture Studies 18:359. Cheng, D.C-H., Hunt, J.A., and Madhvi, P. 1985. Status report on process control viscomters: current applica- tions and future needs. Warren Spring Laboratory, Stevenage, England. Collins, E.A. 1987. private communication. Rheology con- sultant to Mitech Corporation, Willoughby, OH. Cutler, A.N., Morris, E.R., and Taylor, L.J. 1983. Oral perception of viscosity in fluid foods and model sys- tems. J. Texture Studies 14:377. Delancy, M. 1988. private communication. Manager of Tech- nical Support, Haake Division of Fisons Instruments, Saddle Brook, NJ. Dziedzic, 8.2., and Kearsley, M.W. 1984. Physico-chemical properties of glucose syrups. In: "Glucose Syrups: Science and Technology," 8.2. Dziedzic and M.W. Kears- ley, eds., Elsevier Academic Press, London. 90 Gill, J.L. 1978. "Design and Analysis of Experiments in the Animal and Medical Sciences. The Iowa State University Press, Ames. Hardy, R.C. 1962. "NBS Viscometer Calibrating Liquids and Capillary Tube Viscometers." Natl. Bur. Stand. (U.S.) Monograph 55. and Bollenback, G.N. 1966. "This is Liquid Yonkers, NY. Hoynak, P.X., Refined Syrups and Sugars, Inc. 1977. H Sugar. 103 (International Organization for Standardization). Technical Report 3666. J.M., and Thomas, A. 1987. Fluid Jones, T.E.R, Davies, inertia effects on a controlled stress rheometer in its "Carrimed CS500 Operations Manu- In: England. oscillation mode. al," Carrimed, Ltd., H. 1973. Dorking, W., and Pancoast, "Handbook of Sugars." AVI Junk, Publishing Co., Westport, CT. Kawata, M., Kurase, K., and Yoshida, K. 1969. Realization of a viscosity standard. Proc. Fifth Int. Congress on Rheology, Vol I, S. Onogi, ed. p 453. Kreiger, I.M. 1968. Shear rate in the couette viscometer. J. Rheol. 12:5. G.A. 1959. Absolute viscosity of water at a Nauchn-Issled, Inst. Mailiarov, temperature of 20°C. Trudy Vses. Metrl. 37(97):125. private communication. State College, PA. Manning, R.E. 1987. Principal Scien- tist, Cannon Instruments, 1980. Recommended reference materials for section Marsh, K.N., ed. realization of physicochemical properties: viscosity. Pur. and Appl. Chem. 52:2393. The accuracy of measurements of viscosity of liquids. Stds. 75A:535. R.S. 1972. Calibration of viscometers. In: "Pre- R.L. Bloss and M. Marvin, cision Measurement and Calibration", Nat. Bur. Std. (U.S.) Spec. Publ. 300, R.S. 1971. Marvin, J. Res. Nat. Bur. Orloski, Eds., 8:557. 1984. Viscosity of sugar 11:209. "Statistics for Ana- M., and Kasprzyk, P. solutions. Sugar Tech. Rev. Miller, J.C., and Miller, J.N. 1984. lytical Chemistry." Halsted Press, New York. Mathlouthi, 91 Penn, H.W., and Kearsley, E.A. 1971. An absolute determi- nation of viscosity using channel flow. J. Res. Nat. Bur. Stand. (U.S.), 75A(6):553. Porter, R.S., and Johnson, J.F. 1960. Bulk and solution flow properties of polybutenes. ACS Div. Polymer Chem. 1, No. 1. Prentice, J.H. 1984. "Measurements in the Rheology of Food- stuffs." Applied Science Publishers, London. Prilutski, G., Gupta, R.K., Sridhar, T., and Ryan, M.E. 1983. Model viscoelastic fluids. J. Non-Newt. Fluid Mech. 12:233. Pucinich, S. 1989. private communication. Sales agent, Silicon Division, General Electric Corporation, Water- ford, NY. Quagliano, J.V., and Vallarino, L.M. 1969. "Chemistry," 3rd ed. Prentice-Hall, Englewood Cliffs, NJ. Rao, M.A., and Cooley, H.J. 1984. Determination of effective shear rates in rotational viscometers with complex geometries. J. Texture Studies 15:327. Rao, V.N.M, Hamann, D.D., and Humphries, E.G. 1975. Flow behavior of sweet potatoe puree and its relation to mouthfeel quality. J. Texture Studies 6:197. Roscoe, R., and Bainbridge, W. 1958. Viscosity determination by the oscillating vessel method. Proc. Physical Soci- ety 72(4):585. Sanderson, G.R. 1981. Polysaccharides in food. Food Technol. 35(7):50. Simpson, B. 1989. private communication. Purchasing Agent, Kellogg Company, Battle Creek, MI. Skoog, D.A., and West, M.W. 1971. Principles of instrumental analysis. Holt, Rinehart and Winston, New York. Stecher, P.G. 1960. "The Merck Index of Chemicals and Drugs," 7th Edition. Merck and Co., Rahway, NJ. Steffe, J.F., Mohamed, 1.0., and Ford, E.W. 1983. Rheolog- ical properties of fluid foods. ASAE 83-6512. Steffe, J.F. 1987. private communication. Professor of Food Engineering, Michigan State University, East Lansing, MI. 92 Swindells, J.F., Coe, J.R., and Godfrey, T.B. 1952. Absolute viscosity of water at 20°C. J. Res. Nat. Bur. Std. 48(1):1. Szczesniak, A.S., and Farkas, E. 1962. Objective charac- terization of the mouthfeel of gum solutions. J. Food Sci. 27:381. Tung, M.A. 1987. private communication. Chairman, Food Science Department, Technical University of Nova Sco- tia, Halifax, NS. Ubbelohde, L. 1936. The suspended lever viscometer. J. Inst Petroleum Technologists 22:37. United States Government 1977. Legal definition of glucose syrup. Federal Register 42:14479. Uriano, G.A, and Gravatt, C.C. 1977. The role of reference materials and reference methods in chemical analysis. CRC Crit. Rev. in Anal. Chem. 6(4):361. Walker, B. 1989. private communication. Brookfield Engi- neering Laboratories, Stoughton, MA. Weast, R.C. 1971. "Handbook of Chemistry and Physics." The Chemical Rubber Co., Cleveland. Wernimont, G.T. 1985. "Use of Statistics to Develop and Evaluate Analytical Methods." Association of Official Analytical Chemists, Arlington, VA (USA). White, H.S., and Kearsley, E.A. 1971. An absolute determi- nation of viscosity using a torsional pendulum. J. Res. Nat. Bur. Stand. (U.S.), 75A(6):541. Whorlow, R.W. 1980. "Rheological Techniques." Ellis Horwood Limited, Chichester; Wiley, New York. APPENDICES APPENDIX I APPENDIX I. ORIGINAL HAAKE CONCENTRIC CYLINDERS DATA " unprocessed data from an individual The original "raw, Haake analysis consisted of several hundred data points, collected every 2 s, for the typical 10 minute analysis time. Each point in an analysis consisted of time of analy- ‘sis (s), rotor rotational measurement (rpm), and torque measurement (% full scale). This raw data was then con- verted to shear stress and shear rate data, either for each point (for continuous-ramp analyses) or for an average of the last 8 points in each 15-point step (for step-ramp analyses). The shear stress and shear rate data was then fit to several rheological models: Herschel-Bulkley, Bing- ham, and Newtonian. Since the volume of "raw" and "converted" data, for Haake and both Carrimed analysis modes, stretched to several hundred pages, only summaries are presented for all analy- ses. After the data summary, sample sets of raw and con- verted data sets are included as examples. In the table, a position occupied by the < sign means that the entry in that position is the same as for the previous analysis. The Haake was calibrated by removing the sensor head from its holding bracket, laying it on its side, and hanging a series of balance calibration weights from a rotor at- 93 l o‘ ' lel-I'I' “I "LI-II” 94 tached to the head. The weights were suspended from the rotor with thin nylon line, to generate a known torque. The resulting percent full scale torque readings taken from the RV-12 were linearly regressed against the torques. The resulting linear coefficients of slope (units: [N m] / [% full scale torque reading]) and intercept (units: N m) were subsequently used to calculate measured torque directly from observed percent full scale torque readings during fluid analysis. The total number of Haake analyses conducted was 70. The original Haake data is contained in Table 11. 95 Table 11. Original Haake data. Analysis number: 1 2 Parameter: ---------- Sample: oil or syrup oil < Sample: i.d. number $30000 < Sample: lot number 87101 < Sample: load status 30 days < Replicate number 1 2 Avg. temperature, 'C 26.0 Haake sensor head M500 < Haake rotor/cup SVII < Sensor calib. slope x 104, N m/% F.S. rdg. 4.728 < Senso calib. intcpt. x 10 , N m 7.218 < Step or Cont. ramp C < analysis time (min) 5 < max. rotor speed, rpm 64 < Results: maximum 4, s"1 57.6 < Up-ramp fl, Pa 3 61.987 61.777 r2 0.997 0.998 Down-ramp n, Pa 3 62.646 62.860 r2 0.995 0.997 Analysis number: 6 7 Parameter: ---------- Sample: oil or syrup oil < Sample: i.d. number < < Sample: lot number < < Sample: load status < fresh Replicate number 3 1 Avg. temperature, 'C 25.4 25.6 Haake rotor/cup < < Haake sensor head < < Sensor calib. slope x 104, N m/% F.S. rdg. < < Sensor calib. intcpt. x 104, N m < < Step or Cont. ramp < < analysis time (min) < < max. rotor speed, rpm < < Results: maximum 7, s'1 54.0 54.1 Up-ramp n, Pa 3 72.147 70.243 r2 0.997 0.998 Down-ramp fl, Pa 3 72.946 71.275 r2 0.998 0.998 AAAA < 62.016 0.998 63.048 0.996 AAAA 53.9 69.636 0.998 70.630 0.998 4.683 -0.890 < < 60 54.0 73.760 0.997 74.842 0.996 AAAA 53.9 69.128 0.998 70.087 0.998 A AAANAAAA AAAA 53.9 72.762 0.998 73.503 0.998 592.4 0.109 0.999 0.111 0.999 Table 11 (cont’d.). Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Avg. temperature, 'C Haake sensor head Haake rotor/cup Sensoz calib. slope x 10 , N m/% F.S. rdg. Sensor calib. intcpt. x 104, N m Step or Sont. ramp analysis time (min) max. rotor speed, rpm Results: maximum 4, s—1 Up-ramp n, Pa 3 r2 Down-ramp n, 1.2 Pa s Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Avg. temperature, 'C Haake rotor/cup Haake sensor head Senso calib. slope x 10 , N m/% F.S. rdg. Sensor calib. intcpt. x 104, N m Step or Sont. ramp analysis time (min) max. rotor speed, rpm Results: maximum 7, s"1 Up-ramp fl, Pa 3 r30.987 Down-ramp 0, 1.2 Pa 3 96 oil 830000 87101 fresh 1 24.0 M500 SVII 4.893 -1.080 S 5 55 49.5 76.833 0.999 79.667 0.996 AAAA 185.9 19.271 0.989 19.147 0.995 AAAA 49.5 76.833 0.999 79.667 0.997 AAAA 185.6 19.026 0.999 18.945 0.995 AAAA 49.5 75.911 0.999 77.151 0.999 AAA 50 44.2 20.296 0.999 20.665 0.999 14 15 < < 88000 < 86104 < fresh < 1 2 25.0 25.1 < < < < < < < < < < < < 150 200 136.1 185.6 20.574 19.603 0.997 0.987 20.836 19.404 0.997 0.995 19 20 < oil < S600 < 87202 < fresh 6 1 25.0 25.1 < MVA < M500 < < < < < < < < < 258 44.2 233.8 20.194 1.385 0.999 0.999 20.851 1.417 0.999 0.998 Table 11 (cont’d.). Analysis number: 21 Parameter: ----- Sample: oil or syrup oil Sample: i.d. number 3600 Sample: lot number 87202 Sample: load status per 20 Replicate number 2 Avg. temperature, 'C 25.1 Haake sensor head M500 Haake rotor/cup MVA Sensoi calib. slope x 10 , N m/% F.S. rdg. 4.893 Sensor calib. intcpt. x 104, N m -1.080 Step or Sont. ramp C analysis time (min) 5 max. rotor speed, rpm 250 Results: maximum 4, s'1 228.8 Up-ramp n, Pa 5 1.376 r2 0.999 Down-ramp 0, Pa 3 1.412 r2 0.999 Analysis number: 26 Parameter: ----- Sample: oil or syrup < Sample: i.d. number S60 Sample: lot number 87302 Sample: load status fresh Replicate number 1 Avg. temperature, °C 25.0 Haake rotor/cup MVB Haake sensor head M150 Senso: calib. slope x 10 , N m/% F.S. rdg. 1.397 Sensoi calib. intcpt. x 10 N m -0.085 Step or Sont. ramp C analysis time (min) 5 max. rotor speed, rpm 250 Results: maximum 7, s"1 596.4 Up-ramp 0, Pa 3 0.0977 r2 0.999 Down-ramp n, Pa 3 0.101 r2 0.999 AAAA 229.3 1.375 0.999 1.411 0.999 AAAA 596.6 0.097 0.999 0.101 0.999 AAA 214.4 1.379 0.999 1.439 0.999 AAAA 596.6 0.0977 0.999 0.101 0.999 AAAA 214.3 1.365 0.999 1.415 0.999 < < < 200 478.4 0.173 0.999 0.180 0.999 AAAA 214.6 1.356 0.999 1.407 0.999 AAAA 478.0 0.175 0.999 0.184 0.999 Table 11 (cont’d.). Analysis number: 31 Parameter: ----- Sample: oil or syrup syrup Sample: i.d. number 18100 Sample: lot number 7K1X2 Sample: load status per 30 Replicate number 2 Avg. temperature, 'C 25.1 Haake sensor head M150 Haake rotor/cup MVB Sensoi calib. slope x 10 , N m/% F.S. rdg. 1.397 Senso calib. intcpt. x 10 , N m -0.085 Step or Sent. ramp C analysis time (min) 5 max. rotor speed, rpm 200 Results: maximum 7, s' 478.3 Up-ramp n, Pa 3 0.177 r3 0.999 Down-ramp n, Pa 3 0.183 r2 0.999 Analysis number: 36 Parameter: ----- Sample: oil or syrup < Sample: i.d. number < Sample: lot number < Sample: load status < Replicate number 2 Avg. temperature, 'C 25.0 Haake rotor/cup < Haake sensor head < Senso calib. slope x 10 , N m/% F.S. rdg. < Sensoi calib. intcpt. x 10 , N m < Step or Sont. ramp < analysis time (min) < max. rotor speed, rpm < Results: maximum 4, 3'1 596.8 0.096 Up-ramp n, Pa 3 r3 0.999 Down-ramp n, Pa 3 -- r3 --- 98 AAAA 477.6 0.177 0.999 0.184 0.999 478.1 0.179 0.998 596.1 0.095 0.999 0.098 0.999 AAAA 478.3 0.178 0.994 AAAA 597.1 0.095 0.999 0.098 0.999 A AAOA 478.2 0.178 0.999 AAAA 478.4 0.179 0.999 99 Table 11 (cont’d.). Analysis number: 41 Parameter: ----- Sample: oil or syrup syrup Sample: i.d. number 185500 Sample: lot number K720169 Sample: load status fresh Replicate number 1 Avg. temperature, 'C 25.0 Haake sensor head M150 Haake rotor/cup MVB Sensoi calib. slope x 10 N m/% F.S. rdg. 1.387 Sensoz: calib. intcpt. x 10 -0.418 Step or Sont. ramp 8 analysis time (min) 10 max. rotor speed, rpm 50 Results: maximum 4, s' 119.5 Up-ramp n, Pa 3 0.791 r2 0.999 Analysis number: 46 Parameter: ----- Sample: oil or syrup < Sample: i.d. number < Sample: lot number < Sample: load status < Replicate number 2 Avg. temperature, °C 25.1 Haake rotor/cup < Haake sensor head < Sensoi calib. slope x 10 Nm/% F.S. rdg. < Sensoi:ca1ib. intcpt. x 10 < Step or Sont. ramp < analysis time (min) < max. rotor speed, rpm < Results: maximum 1, 8.1 359.3 Up-ramp fl, Pa 8 0.825 r3 0.999 oil 8600 87202 fresh 1 25.2 M500 MVB 4.920 -2.602 < 8.5 95 226.5 1.335 0.999 AAAA 322.7 0.812 0.999 AAAA 359.3 0.815 0.999 < < < < 226.7 1.369 0.999 7350 7K11X105 fresh 1 25.0 SVII < 4.893 0.437 < < 85 76.6 51.87 0.997 syrup 185500 K70169 fresh 1 25.0 < < < < 10 150 359.3 0.829 0.999 AAAA 76.8 51.80 0.997 100 Table 11 (cont’d.). Analysis number: 51 Parameter: ----- Sample: oil or syrup syrup Sample: i.d. number 7350 oil 830000 Sample: lot number 7K11X105 87101 Sample: load status per 49 Replicate number 3 Avg. temperature, °C 25.1 Haake sensor head M500 Haake rotor/cup SVII Sensoi calib. slope x 10 , N m/% F.S. rdg. 4.893 Sensoi calib. intcpt. x 10 , N m 0.437 Step or Sont. ramp 8 analysis time (min) 10 max. rotor speed, rpm 85 Results: maximum 7, s” 76.6 Up-ramp fl, Pa 8 51.84 r2 0.997 Analysis number: 56 Parameter: ----- Sample: oil or syrup < Sample: i.d. number 1300 Sample: lot number 7L11260 Sample: load status fresh Replicate number 1 Avg. temperature, 'C 25.1 Haake rotor/cup < Haake sensor head < Senso calib. slope x 10 , N m/% F.S. rdg. < Sensoi calib. intcpt. x 10 , N m < Step or Sont. ramp < analysis time (min) < max. rotor speed, rpm 50 Results: maximum 7, s'1 44.9 Up-ramp n, Pa 3 82.67 2 r 0.999 fresh 1 25.0 < < < < < < 60 54.5 78.05 0.997 AAAA 45.0 83.70 0.999 AAAA 45.0 84.64 0.999 syrup 7350 7K11X105 fresh 1 25.0 < < < ( < < 85 76.0 47.18 0.997 45.3 87.17 0.999 AAAA 76.2 46.40 0.998 < < < < 45.1 86.63 0.999 101 Table 11 (cont’d.). Analysis number: 61 Parameter: ----- Sample: oil or syrup syrup Sample: i.d. number 4400 Sample: lot number 7J29X312 Sample: load status fresh Replicate number 1 Avg. temperature, °C 25.0 Haake sensor head _ M500 Haake rotor/cup SVII Sensoi calib. slope x 10 , N m/% F.S. rdg. 4.757 Sensor calib. intcpt. x 104, N m -1.236 Step or Sont. ramp 8 analysis time (min) 10 max. rotor speed, rpm 50 Results: maximum y, s-1 45.2 Up-ramp n, Pa 3 86.96 r2 0.999 Analysis number: 66 Parameter: ----- Sample: oil or syrup syrup Sample: i.d. number 7350 Sample: lot number 7K11X105 Sample: load status fresh Replicate number 1 Avg. temperature, °C 25.0 Haake rotor/cup < Haake sensor head SVII Sensoi calib. slope x 10 , N m/% F.S. rdg. < Sensoi calib. intcpt. x 10 , N m < Step or Sont. ramp < analysis time (min) < max. rotor speed, rpm 85 Results: maximum 4, 3'1 76.2 Up-ramp n, Pa 3 53.34 r2 0.998 AAAA 76.5 53.73 0.998 300 7L4X24 fresh 1 24.9 < < < < < < 27 22.5 171.54 0.999 AAAA 76.5 51.52 0.997 64 65 < oil < 830000 < 87101 < fresh 2 1 24.9 25.0 < < < SVI < < < < < < < < < 18 22.6 16.1 127.49 78.65 0.999 0.999 69 70 < < < 1300 < 7L1126 < fresh 2 1 25.0 25.0 < < < < < < < < < < < < < < 76.9 79.0 54.74 53.72 0.997 0.992 APPENDIX II APPENDIX II. ORIGINAL CARRIMED STEADY SHEAR DATA I The original "raw,' unprocessed data from Carrimed analyses were written in IBM-PC computer files, one file per individual analysis. The files were written in ASCII text, in a proprietary format. The files contained sample identi- fication, analysis conditions (temperature, fixtures, analy- sis mode, etc.), programed torques and the measured angular velocities or amplitudes. An option in the Carrimed control and analysis software was used to compute, from the raw data file, shear stress - shear rate values at each sample point (typically 2—3 per second, over an analysis time period of several minutes). The program then fit a variety of shear models to the data set, allowing the user to control which points from the data set to include in the least-squares re- gression computations. The regression coefficients reported in the data summary below were generated using this ap- proach. The shear stress - shear rate curves used in the figures above showing Carrimed data were calculated and printed to a computer disk using a Lotus 123 program. The figures are identical to those generated by a similar utili- ty in the Carrimed software program. 54 Carrimed steady shear analyses results are summa- rized below in Table 12. The units of sensor calibration slope are (N m) / (% full scale reading). 102 Table 12. Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, °C Cone angle, deg-min Cone diameter, cm Cone-plate gap, pm Solvent trap? Maximum torque, pN m Analysis time (min) Results: maximum 1, s-1 Up-ramp 0, Pa 3 r3 Down-ramp 0, r2 Pa 3 Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, °C Cone angle, deg-min Cone diameter, cm Cone-plate gap, pm Solvent trap? Maximum torque, pN m Analysis time (min) Results: maximum 7, s- Up-ramp 0, Pa 3 r2 Down-ramp n, 1.2 Pa 3 103 4 26.1 N 1000 4 3.1 18.80 0.9999 19.07 0.9999 AAAAAWAAAA 1000 < 0.9 20.64 0.9999 20.74 0.9999 3.0 19.63 0.9999 20.10 0.9999 50000 4 1300 17.65 0.9932 Original Carrimed steady shear data. AAAAAA 2.9 19.51 0.9999 20.01 0.9999 40000 < 1100 19.44 0.9972 17.36 0.9972 4 5 < < < < ( fresh fresh 1 2 24.5 < < < < < < < < < 5000 < < < 4.0 4.1 22.03 21.27 0.9999 0.9999 22.35 21.53 0.9999 0.9999 9 10 < < 860 < 86301 < fresh fresh < < 24.5 < 1-0 < 4 < 26.1 < < < 1000 < < < 42 42 1.411 1.406 0.9999 0.9999 1.418 1.413 0.9999 0.9999 Table 12 (cont’d.). Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, ’0 Cone angle, deg-min Cone diameter, cm Cone-plate gap, pm Solvent trap? Maximum torque, uN m Analysis time (min) Results: maximum 4, s- Up-ramp n, Pa 3 r3 Down-ramp 0, r2 Pa 3 Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, °C Cone angle, deg-min Cone diameter, cm Cone-plate gap, um Solvent trap? Maximum torque, uN m Analysis time (min) Results: maximum 7, s- Up-ramp n, Pa 3 r3 Down-ramp n, 1.2 Pa 3 104 11 12 oil syrup S600 I8100 86301 7A27X221 per 10 fresh 2 1 25.0 < 1-0 < 4 < 26.1 < N < 1000 250 4 < 44 77 1.339 0.195 0.9999 0.9997 1.344 0.186 0.9999 0.9997 16 17 oil < 8600 < 86301 < fresh fresh 1 2 < < < < < < < < < < 1000 < < < 47 48 1.254 1.234 0.9999 0.9999 1.257 1.236 0.9999 0.9999 13 14 15 < < < < 4300 < < 7013XLT7 < fresh fresh 3rd 1 1 1 < < < ( < < < < < < < < < < < 5000 < 50000 < < < 1828 8.5 102 0.1608 --- 29.00 0.9999 ..._ 0.9989 0.163 22.53 --- 0.9999 0.9976 --- 18 19 20 < < < < < 88000 < < 86104 fresh fresh fresh 1 2 1 < < < < < < < < < < < < < < < 40000 < 1000 < < < 2200 2200 3 1.122 1.088 18.67 0.9991 0.9989 0.9999 1.108 1.072 19.05 0.9996 0.9999 0.9999 Table 12 (cont’d.). Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, °C Cone angle, deg-min Cone diameter, cm Cone-plate gap, pm Solvent trap? Maximum torque, pN m Analysis time (min) Results: maximum 4, s—1 Up-ramp n, Pa 3 r2 Down-ramp n, 1.2 Pa 3 Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, °C Cone angle, deg-min Cone diameter, cm Cone-plate gap, p Solvent trap? Maximum torque, pN m Analysis time (min) Results: maximum 4, 3'1 Up-ramp 0, Pa 3 r2 Down-ramp 0, 1.2 Pa 3 105 4 26.1 N 50000 4 175 17.60 0.9996 17.50 0.9999 AAAAAHSAAA 1500 < 3 19.41 0.9998 19.91 0.9997 AAAAAI-‘(DAAA 1000 < 3 18.70 0.9999 19.14 0.9999 01 O AOAAAAANAAAA 00 170 17.38 0.9998 17.44 0.9999 01 O AOAAAAAHAAAA O O 175 17.98 0.9997 17.89 0.9999 AAAAAI—‘SAAA 1000 < 2.5 19.79 0.9997 19.79 0.9998 24 25 syrup < 3260 < 212-214 < fresh < 1 2 < < < < < < < < Y < 1000 50000 < < 3.0 186 19.95 16.01 0.9999 0.9996 20.27 16.24 0.9999 0.9998 29 30 oil < 8600 < 86301 < fresh < 1 2 < < < < < < < < N < < < < < 45 45 1.292 1.294 0.9999 0.9999 1.295 1.296 0.9999 0.9999 Table 12 (cont’d.). Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, 'C Cone angle, deg-min Cone diameter, cm Cone-plate gap, pm Solvent trap? Maximum torque, pN m Analysis time (min) Results: maximum 4, s- Up-ramp fl, Pa 3 r2 Down-ramp n, r2 Pa 3 Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, 'C Cone angle, deg-min Cone diameter, cm Cone-plate gap, p Solvent trap? Maximum torque, pN m Analysis time (min) Results: maximum 7, s' Up-ramp n, Pa 3 r2 Down-ramp n, 1.2 Pa 3 106 26.1 N 40000 4 1900 1.234 0.9997 1.221 0.9999 AAAAAAAHCDAAA 35 94.07 0.9999 94.05 0.9998 140 21.04 0.9999 20.94 0.9999 21.23 0.9999 21.68 0.9999 59 1.345 0.9999 1.352 0.9999 AAAAAAAACDAAA 2.8 21.83 0.9999 22.25 0.9999 < 860 87302 2nd 0 AOAAAA 85 0.094 syrup 4400 7J29X FRESH 1 26.1 Y 50000 < 31 92.80 0.9996 0.9997 0.094 94.34 0.9995 0.9998 AAAAAAANI‘DAAA 3.0 21.72 AAAAAAAODAAAA 3.0 21.86 0.9998 0.9998 21.72 21.86 0.9999 0.9998 Table 12 (cont’d.). Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, °C Cone angle, deg-min Cone diameter, cm Cone-plate gap, u Solvent trap? Maximum torque, pN m Analysis time (min) Results: maximum 4, s—1 Up-ramp 0, Pa 3 r2 Down-ramp n, 1.2 Pa 3 Analysis number: Parameter: Sample: oil or syrup Sample: i.d. number Sample: lot number Sample: load status Replicate number Temperature, °C Cone angle, deg-min Cone diameter, cm Cone-plate gap, pm Solvent trap? Maximum torque, pN m Analysis time (min) Results: maximum 4, s- Up-ramp 0, Pa 3 r2 Down-ramp 0, 1.2 Pa 3 107 oil 830000 87101 fresh 1 25.0 1-0 2 28.8 N 41888 < 260 77.82 0.9997 77.10 0.9999 AA