Design and use of a thermoset capillary rheometer - American

oil from a process heat transfer system through a surrounding jacket. ... to a movable frame which was inserted Intoa constant crosshead speed testing...
1 downloads 0 Views 471KB Size
Ind. Eng. Chem. Prod. Res. Dev. 1084, 23, 103-106

103

Design and Use of a Thermoset Capillary Rheometer Salvatore C. Malguarnera;

Douglas R. Carroll,' and Marlo A. Colaluca

Department of Mechnlcel Englneerlng, Texas A&M University, College Station, Texas 77843

Capillary rheometry is an effective means for evaluatlng the processing behavior of plastic molding compounds. However, the use of this technique with thermosets can produce a number of problems. Temperature control is much more critical than with thermoplastics. Hot spots in the barrel or capillary can lead to unwanted local cuing. Clearing the barrel of set up material can also be quite troublesome. With these problems in mind, a capillary rheometer was designed specifically for use with thermosets. The rheometer barrel was heated by circulating oil from a process heat transfer system through a surrounding jacket. The barrel-jacket assembly was attached to a movable frame which was inserted into a constant crosshead speed testing machine. Thls system was used successfully with six commercially available thermoset molding compounds. Standard capillary rheometry procedures were used to determine the dependence of the apparent shear vlscosity of these materials on shear rate, temperature, and heating time.

General Aspects of Thermoset Rheology The flow of any thermosetting material can be separated into two regions of interest by the concept of a gel point. Prior to gelation the resin is a flowable melt, which can generally be described as a shear thinning material, characterized by an apparent shear viscosity. The onset of gelation is signaled by a rapid increase in apparent viscosity. Beyond this point the material is best characterized by dynamic type mechanical measurements and calorimetry (White, 1974; Maas, 1978). Thus before gelation the apparent viscosity of a thermosetting material is a function of time, temperature, and shear rate. Behavior subsequent to gelation, since no flow is involved, depends only on time and temperature. Thermosets exhibit a characteristic U-shaped curve when the apparent viscosity is plotted vs. time or temperature for a fixed shear rate. The downward part of the curve is representative of thermal softening and shear thinning flow. A minimum is reached and then a rapid increase in viscosity caused by the cross-linking reaction is observed. A great deal of information concerning processing behavior can be determined from such viscosity-timetemperature profiles: maximum residence time in the screw barrel, mold curing time, minimum viscosity, and the temperature sensitivity of the material (White, 1974; Roller, 1975). A plot of apparent viscosity as a function of shear rate for different temperatures also provides important information for understanding mold filling. The shear viscosity will also be affected by a number of other considerations. Moisture can greatly reduce the apparent viscosity and lead to blistering and porosity in the molded part (Hess, 1971). Fillers can change the shear viscosity in two ways. The shear thinning behavior prior to gelation can be affected by filler geometry, type, surface treatment, and concentration. The kinetics of cross-linking can be influenced by reactive complexes on the filler's surface (Dutta and Ryan, 1979). Additives can also change the viscosity (Mussatti and Macosko, 1973) and thermal stability (Delvigs et al., 1979). A number of techniques have been used to characterize the flow behavior of thermosets. Among the most popular have been spiral flow molds (Hess, 1971; Heinle and Rodgers, 1969; ANSI/ASTM, 1978), minimum pressure to fill a cavity (Sundstrom and Walters, 1971),Brabender 'Addreas correspondence to this author at 403 Genevieve Drive, Lafayette, LA 70503. *Tinker Air Force Base, OK 73145.

and other type torque rheometers (Choi, 1970; Sundstrom and Walters, 1971; Beck and Golovoy, 1972; Driscoll et al., 1975), cone and plate rheometers (White, 1974; Macosko and Mussatti, 1972; Frizelle and Norfleet, 1971a), injection molding machine operations (Beck and Golovy, 1972; Frizelle and Norfleet, 1971b), and capillary rheometers (Sidi and Kamal, 1980; Slysh and Guyler, 1978). Many of the above techniques require the use of injection molding machines and give at best semiquantitative results. Other methods are restricted to low shear rate measurements which may not give a good indication of the material's behavior in actual processing. A simple and effective means of evaluating thermoset flow behavior which would provide quantitative results directly applicable to process engineering would be most welcome. The capillary rheometry technique is well established, convenient, and the only method able to yield shear viscosity data at the high shear rates characteristic of the injection molding process. However, the use of this technique with thermosets can produce a number of problems. Temperature control is much more critical with thermosets than with thermoplastics. Hot spots in the barrel or capillary can lead to unwanted local curing. Cleaning the barrel of material which has set up can be a major problem. Capillary Rheometer Design Commercially available capillary rheometers have been designed exclusively for use with thermoplastic materials. Electric band heaters are used to maintain the barrel temperature, and a rather limited maximum load is available to force material from the barrel. Therefore these devices are unsuitable for use with thermosets. With these problems in mind a capillary rheometer was designed specifically for use with thermosets. A schematic representation of the system built is shown in Figure 1. The rheometer barrel was heated by circulating Dowtherm G fluid from a process heat transfer system through a surrounding jacket. Temperatures up to 288 OC (550 O F ) could be attained with the heating element used. Higher temperatures up to the fluid's maximum of 343 OC (650 O F ) could be reached with a more powerful heating unit. Temperature control was provided by a proportional temperature regulator monitoring the temperature of the fluid in the barrel jacket. The variation in temperature from the set point was approximately 0.5 "C at 260 "C (500O F ) . Fluid entered the jacket at the bottom and exited near the top, thus ensuring that the barrel would always be covered to the same level. The capillary itself was placed relatively high up in the barrel in order to ensure that its tempera-

0196-4321/84/1223-0103~01.5Q/O @ 1984 American Chemical Society

Ind. Eng. Chem. Prod. Res. Dev., Vol. 23,No. 1, 1984

104

, I

L--_

~

Figure 1. Schematic of thermoset capillary rheometer.

ture would be close to that of the surrounding fluid. Several samples which cured in the barrel showed smooth surfaces, indicating that the temperature was uniform along the barrel. The barrel-jacket assembly was attached to a movable frame which was inserted into a constant crosshead speed testing machine. Flexible stainless steel hoses allowed the barrel-jacket assembly to be separated from the heat transfer system. The testing machine used could provide up to 20 000 lb of force. This allowed any material which set up in the barrel to be pushed out by the piston, since its length of travel could extend past the capillary position in the barrel. Any material in the capillary was removed by the use of a thin steel rod. A number of specific points are worthy of mention. In order to avoid leaks, especially vapor leaks, good sealing is necessary at all joints. Only metal to metal seals should be considered. Silicone and PTFE were initially used to seal the capillary barrel to the heating jacket. These seals proved unsuitable after limited use and the barrel was welded to the jacket and no further problems were encountered. For added safety a well-ventilated work area should be used. Finding an adequate high-temperature pump for the heat transfer system proved exceedingly difficult. Only a few manufacturers carry the low capacity, low-pressure pump required. Although these were stock pumps specifically modified for high temperature operation, the one selected gave considerable problems until additional internal clearances were cut. Manufacturer claims should be viewed as being optimistic at best. The reservoir and the rheometer heating jacket, as well as all the fluid lines must be well insulated with hightemperature insulation material. The heat exchanger unit used allowed direct contact of the heating fluid and the electrical resistance heaters. Thus, the maximum allowable watt density of the heat transfer fluid selected will control the maximum power of the heat exchanger. This requirement is necessary in order to avoid decomposition of the heat transfer fluid. If additional heating capacity is desired in order to decrease start-up time, a physically larger unit must be used. By placing the heat exchanger, pump, controller, and reservoir on a movable cart, the system can be used with any available testing machine. This will also facilitate storage and transportation. A filter to remove foreign matter from the fluid is recommended. It can be best placed on the return port to

the reservoir since this location will make it easier to remove and clean. A simple wire type element will do. Experimental Procedures Standard techniques for capillary rheometry were used (Bird et al., 1977; Whorlow, 1980). The barrel was first heated with the capillary and capillary support in place. The temperature was monitored by placing a thermocouple down the barrel. When the temperature had stabilized, the piston was inserted and allowed to heat for 15 min. The sample material was then poured into the barrel and compressed by the machine until the load increased sharply. The sample was heated according to the experimental program. Tests were then conducted with the piston descending at constant speed. The resulting force on the load cell of the testing machine was automatically plotted as a function of time. The apparent viscosity 7 is given as 7 = ~R/?R

where 7 R is the shear stress and qRis the shear rate at the capillary wall. Ignoring entrance and exit effects T R = hpR/2L where AP is the pressure drop across the capillary, R is the capillary's radius, and L is its length. The shear rate at the wall is given as

where Q is the volumetric flow rate. Experimental Results Initial testing was conducted with thermoplastic materials in order to check the equipment and experimental techniques. Data taken were reproducible. Results obtained using the 2.54-cm (1 in.) capillary agreed closely with results obtained using the 3.81-cm (1.5 in.) capillary. Few data are available in the literature to compare the results obtained by this work. Frizelle and Norfleet (1971b) used a commercially available capillary rheometer with a small L f D ratio of 3.8 to study two phenolic compounds. The phenolic compound studied in this investigation showed a viscosity in the same order as that of Frizelle and Norfleet, averaging between one-half and two-thirds their reported values. Six thermoset molding compounds were investigated: an alkyd polyester, a diallyl phthalate (DAP), a phenolic, a woodflour-filled phenolic, a mineral filled alkyd polyester, and a woodflour-filled melamine. The instrument functioned satisfactorially with all these materials. In the interest of brevity only the results obtained with the woodflour-filled phenolic will be discussed since this material's behavior is typical of thermoset molding compounds. Details of the other materials performance are available (Carroll, 1982). A capillary diameter of 0.16 cm and length of 3.94 cm, LID = 24, was used in all the tests discussed below. The apparent viscosity of the woodflour-filled phenolic as a function of shear rate at 121 OC (250 OF) and different heating times is shown in Figure 2. A power law type behavior is generally displayed. The effect of heating time at a barrel temperature of 121 "C (250 O F ) and different shear rates is shown in Figure 3. The characteristic Ushaped curve has a minimum in the region of 1.5 t~ 2 min. At times greater than 4 min cross-linkingwas great enough to prevent flow. A similar behavior of viscosity with temperature at a fixed heating time of 3 min for different shear

Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984 100.00

105

100,000

9

i

1

1,000 10

%

' 160

' '

"iboo

'

Shear R o t e , sec-l

Figure 2. Effect of shear rate on the apparent viscosity of a woodflour-filled phenolic at 121 OC (250 OF). Legend for heating times: (0) 1 min; (0) 3 min. 100,000

i

,

1

l

1

O

1 ,oo 100

105

110

115

125

120

Temperature,

130

C*

Figure 4. Apparent viscosity as a function of temperature for a woodflour-filledphenolic with a heating time of 3 min. Legend for shear rates is the same as in Figure 3. lO0,OO

l0,OOOj

:

1,000

H e a t i n g T i r e , minutes

Figure 3. Apparent viscosity as a function of heating time for a woodflour-fiiedphenolic at 121 O C (250 OF). Legend for shear rates: (0)81.4 8-'; (A)203.2 s-'; (0) 684.2 8-l; ( 0 )1535.4 s-'.

rates is shown in Figure 4. The shear rate in both Figures 3 and 4 seems to affect the magnitude of the viscosity but not the nature of ita behavior with time or temperature. By plotting the viscosity against the inverse of absolute temperature for a fixed heating time and shear rate an apparent activation energy for viscous flow at constant shear rate can be determined. The data of Figure 5 were obtained for a heating time of 1min. The activation energy for flow of this material shows little shear rate dependence and has an average value of 8.23 kcal/mol. Heat transfer to the granular mass is governed by the contact thermal resistance at the barrel surface. Although thermal gradients between the outer and inner surfaces of the granular mass were not larger, the order of 5 to 10 O F , the temperature of the mass tended to lag that of the barrel by a considerable amount. Thus the temperatures cited should be considered as nominal temperatures for the material, Conclusions Capillary rheometry can be effectively used with thermosets if the equipment employed is designed to overcome

1.00 136

138

1.40

1.42

1.44

1.46

in, *R-'

Figure 5. Apparent viscosity as a function of inverse temperature for a woodflour-filled phenolic with a heating time of 1 min. Legend for shear rates is the same as in Figure 3.

some of the inherent problems of these materials. Most important is providing a uniform temperature in the barrel. This has been successfully accomplished by using a circulating hot oil system. Standard capillary rheometry methods can provide the dependence of thermoset apparent viscosity on shear rate, temperature, and time. Such information is very useful in evaluating resin processability and in setting preliminary fabrication conditions. Acknowledgment The authors wish to thank NASA, Lewis Research Center, for their sponsorship of this work and especially Dr. T. T. Serafini and Mr. R. D. Vannucci for their interest and help; Professor M. E. Ryan of SUNY,Buffalo, for so

106

Ind. Eng. Chem. Prod. Res. Dev. 1904, 23, 106-111

freely discussing his experiences with thermoset rheometry; Mr. L. L. Earles for his great efforts in making the heat transfer system; and Messrs. S. Stanley and A. Leach for their help at various stages in the experimental program. Literature Cited ANSIlASTM 03123-72 (Reapproved 1978) "Standard Test Method for Spiral Flow of Low-Pressure Thermosetting Molding Compounds". Beck, R. H., Jr.; Golovoy, A. S f f A M E C Tech. Pap. 1072, 78, 87-92. Bird, R. B.; Armstrong, R. C.; Hassager, 0. "Dynamics of Polymerlc Liqulds": Vol. 1; Wiley: New York, 1977. Choi, S. Y. SPE J. 1070, 26, 51-4. Carroll, D. R. M. S. Thesis, Texas A 8 M. University, 1982. Debigs, P.; Alston, W. B.; Vanucci, R. D. NASA TM 79062, 1979. Driscoii, S.; Zlmbone, P.; Dubreull, M.; Egber, R.; Alvarez, R. S E A N E C Tech. Pap. 1075, 2 0 , 99-101.

Dutta, A.; Ryan, M. E. J. Appl. Polym. Sci. 1070, 2 4 , 635-49. Frireile, W. G.; Norfleet, J. S. Sf€ANT€C Tech. Pap. 1071, 77, 397-402. Frirelle, W. G.; Norfleet, J. S. SPE J. 1071, 27, 44-8. Heinle, P. J.; Rodgers, M. A. SPE J. 1080, 25, 56-61. Hess, J. E. Mod. Plast. Nov 1071, 60-1. Maas, T. A. M. M. Po&" f n g . Sci. 1078, 78, 29-31. Macosko, C. M.; Mussatti, F. G. SPEANTEC Tech. Pap. 1072, 78, 73-80. Mussatti, F. 0.; Macosko, C. W. Polym. Eng. Sci. 1073, 73, 236-40. Roller, M. B. Polym. Eng. Scl. 1075, 75, 408-14. SMI, S.; Kamal, M. R. SPf A N E C Tech. Pap. 1080, 26, 86-94, Slysh, R.; Guyler, K. Polym. Eng. Sci. 1078. 78, 607-10. Sundstrom, D. W.; Walters, L. A. S E J. 1071, 27, 58-62. White, R. P. Polym. Eng. Sci. 1074, 74, 50-7. Whoriow, R. W. "Rheologlcai Technlques", Ellis Horwood: Chichester, 1980.

Receiued for review April 27, 1983 Accepted September 19, 1983

Pyrolysis Feedstock Studies: Correlation and Modeling of Hydrogenated Distillates Raymond J. Rennard' and Harold E. Swift Gulf Research & Development Company, Pittsburgh. Pennsylvania 75230

This paper reports the effect of l i i t gas oil hydrogenationon pyrolysis yields. Hydrogenation, in addiin to removing sulfur, Significantly aiters product yields upon pyrotysls. Hydrogenation results in significantly increased olefin, fuel gas, and aromatic yields and drastically reduces fuel oil yields. By implication, this should lead to reduced fouling and longer run lengths. All of the changes in product yields can be predicted on the basis of changes In feedstock composition and physical properties resulting from hydrogenation. By use of these parameters, a model has been developed which accurately predicts all product yields for hydrogenated distillates.

Introduction While the use of distillate as an olefin feedstock in the United States has fallen far short of earlier projections (Swift et al., 1978), the current crude oil surplus, with its resultant lower crude prices, combined with rising LPG costs has resulted in a renewed interest in cracking liquid feedstocks. Cost effective methods of increasing olefin yields from distillates and utilization of lower cost sour distillates would significantly increase the incentive for light gas oil cracking. Because of their high aromatic and naphthenic content, the cracking of light gas oils results in significantly lower olefin yields and high yields of pyrolysis fuel oil. Since feedstock sulfur is mostly contained in the refractory aromatics fraction, sulfur is concentrated in the fuel oil product during pyrolysis. This environmentally limits the use of all but very low sulfur feedstocks. Over the years a variety of schemes have been investigated for upgrading distillate range olefin feedstocks. By far the most attractive of these schemes is hydrotreating. Hydrotreating allows a wider selection of feedstocks in that gas oils derived from high sulfur crudes can be cracked in conventional plants without violating environmental restrictions. In addition, the reduction in aromatic content and the increase in feedstock hydrogen content should result in increases in olefins yields and significant decreases in pyrolysis fuel oil formation. This has been experimentally verified, and the economic benefits of gas oil hydrogenation have been reported by Rennard and Swift (1981). The present paper summarizes the trends and 0196-4321 18411223-0106$01.50/0

Table I. Pyrolysis Pilot Plant Operating Conditions gas oil feed rate dilution steam rate steam/oil radiant coil inlet coil outlet average P coil outlet temperature transfer line quench RPG range fuel oil

4.5 kg/h 3.375 kg/h 0.75 41 psig 1 5 psig 26 psig 760-805 "C 538 "C C5 's-3 7 5 F 375 "F

Table 11. Hydrogenation Pilot Plant Operating Conditions hydrogen pressure reactor temperature LHSV hydrogen rate aromatic saturation sulfur removal

600-2500 psig 300-374 "C 1-4 h-' 7000-10 000 scf/bbl 17-95% 97-100%

correlations observed in cracking hydrogenated gas oils and discusses the parameters which were found to be significant in developing yield models for cracking both raw and hydrogenated gas oils. Experimental Section All pyrolysis experiments were carried out in a one barrel-per-day,continuous operation pilot plant. This pilot plant was operated for Gulf Oil Chemicals Co. by Gulf Research & Development Co. The pilot unit was designed to simulate a modern naphtha/gas oil cracker in terms of both product yields and operating conditions. Yields ob@ 1984 American Chemical Society