Harwood
Engineering Company, Inc.
Harwood
Engineering Company, Inc.
This paper describes, from an historical view, the evolution of static and dynamic pressure seals based on the unsupported area and the controlled-clearance principles. Further, the paper gives a quick overview of the equipment made possible by these seals as used in the production of polyethylene, diamonds, hot and cold isostatic compactions etc. The ongoing basic high pressure research in physics, geo-physics, geochemistry, geology, petrology, chemistry, pharmaceuticals, and presently biology was made possible by the development of these reliable and simple packings and seals.
The need to use high pressure as a research tool has led, by necessity, to a continual, urgent effort in the studies of thermodynamics and the properties of fluids under pressure. This is an ongoing effort attested by our current interest and activity in pressure transmissibility, fluid viscosity and speed of pressure transmission. While not discussed here, concurrent evolution in the qualities and varieties of steels and other materials of construction and the high technology of electronics and computers have had significant contributions to today's high pressure capabilities.
Amagat5, a 19th century high pressure investigator, worked at approximately 4 kilobar (60,000 psi) using the packing shown in Figure 3(a). It takes advantage of the low modulus and large Poisson's ratio of the soft annular packing shown in Figure 3(a). At higher pressure the soft packing would extrude into the crevice (e) which is at atmospheric pressure. At Harwood, considerably higher pressures were reached by preventing that extrusion with anti-extrusion wedge rings shown in Figure 3(b). Note that the upper ring must stand a greater load than the lower ring since the system pressure counteracts the extrusion force in the lower ring. The mechanical force (F) must more than balance the force created by the system pressure (P) acting on the bottom face of the closure (a).
The packing shown in Figure 4 also takes advantage of the low modulus and large Poisson's ratio of the soft packing. (a) is backed by the anti-extrusion wedge ring (d) and a reasonably fitted hardened steel plug (c), all held in place by the screw (b). This packing was originally devised at Harwood during some test work; the original soft packing "e" was an eraser from a new wooden pencil.
Professor P.W. Bridgman of Harvard University, Nobel Laureate, famous for his work in the Physics of High Pressure, found it necessary to design his own experimental equipment, including static and dynamic seals for fluids in excess of 35 kilobar (500,000 psi). His original recognition of the principle of "unsupported area" which he used extensively and in many variations is illustrated in a typical dynamic seal, Figure 5, and static seal, Figure 6, of his design.6
In Figure 5 the mushroom shaped piece (a) is free to move in the socket piece (e) with variations of pressure. The area of its stem is the unsupported area. The soft packing (c) is squeezed in the annulus between the socket piece (e) and the mushroom head (a) to pressure slightly in excess of the system pressure (P). The triangular shaped cross sectional pieces of bronze (b) move to prevent extrusion of the soft packing. The packing is self-sealing to approximately 30 kilobar.
The static closure packings, Figure 6, are carried by the closure piece (a) as shown enlarged at one side. The ring (d) has a square cross section and rests upon the closure slope leaving a void space (h). It is followed by the lead ring (e) and extractor piece (f). As Bridgman used it, the extractor rests on a shoulder (g) machined into a vessel. The closure nut (b) squeezes the seal rings against the shoulder making them expand against the vessel wall. The initial seal is made with the lead ring (e), as the system pressure increases, the pressure on the soft steel ring (d) makes an increasingly better seal, thus preventing the lead seal from extruding. The void space (h) provides the unsupported area in this type of seal.
Unsupported area seals have in common a geometry such that as the pressure load increases, the force effecting the seal increases in proportion to the system pressure, making a tighter and tighter seal. All lip type packings must make an initial seal in order to function. While an excessive pumping rate will occasionally appear to seal a small leak, the seal nevertheless is not working properly; usually one must anticipate major seal trouble.
The ring (b) and the test sample are compressed simultaneously by the force (F) developing pressure (P1) between the anvil and ring faces and at the same time generating pressure (P2) in the sample which, acting on the inner diameter of the ring, tends to overcome the friction between the ring and anvils. Obviously, in a given design the thinner the ring (b), the greater the pressure P2 that can be developed. The extra supporting ring (d), the "garter", in Figure 8(b) shows another way that the pressure P2 on the sample can be increased. Additional design discussion would be too lengthy here, since the interplay of materials of construction, the relative proportion of components, the various pressure and temperature techniques are quite straight forward problems.
The anvil can be made from spheres9,10 that are transparent, such as sapphire or glass. Metallic spherical anvils have not been used routinely to 50 kilobars in chemical experiments without recourse to the support garter shown in Figure 8(b). The garter and anvil materials must be selected with compatible properties of compression strength and ductility.
Two types of metal to metal seals are shown: The usual high pressure fitting joints, Figure 9, which depend for their functioning on the force generated by the mechanical screw, and those that make use of unsupported area (Figure 10(a & b), and Figure 11). The screw load makes a large contact stress on the small area of the seat in the fittings.11 For the seal to work, the contact stress must be greater than the process pressure. Note that these fitting seals are not unsupported area seals.
The other metal to metal seals make use of the unsupported area principle as well as using the technique of inducing a large intensity of force on a relatively small seat. This technique is illustrated by the ball check valve Figure 10(a), where the ball sits on a seat of small area, and has an unsupported area the size of the hole in the lens ring.
The optical flat valve, Figure 10(b), depends for its initial seal on the mating of two virtually perfectly flat surfaces; and, after the initial seal is made, the unsupported area makes the seal as in the ball check valve.
This same technique is employed for sealing windows in pressure vessels. A closure showing a window used in a vessel for 14 kilobar service at 4°K is shown in Figure 11. In this application, where the seal needs only to be made up once, the initial seal is "encouraged" by the use of a film of Canadian balsam or other soft material between the optically flatted surfaces to accommodate any minor imperfections in the surfaces. This illustration also shows an electrical leadthrough. It is another application of the unsupported area in combination with the wedge, and was first used by Amagat in the 19th century.12
In the early 1950's Harwood invented the controlled-clearance principle as the basis of design of a new type of free piston primary pressure standards gage.13,14 The controlled clearance free piston gage has since become the primary pressure standard at the National Institute of Standard and Technology (formerly the United States Bureau of Standards) for pressures above the levels encountered with the mercury columns.15 The controlled clearance gage now forms the basis for pressure standards in the United States, in much of Western Europe, and Japan.16 It has been used over a pressure range of a few inches of water through 30 kilobar.17 So much has been written about the controlled clearance gage that more than a brief review here is inappropriate.18
While the first application of the controlled clearance principle was used on primary pressure standard devices, it is readily apparent that the principle can be applied to other equipment19; one example is pressure vessel closures when the vessel is to be subjected to extreme pressure and extreme temperatures and with no leaks to be allowed, where no other seal will work. Illustrative sketches follow. In Figure 12 is shown the principle of controlled clearance; Figure 13 illustrates how it is employed as a packing gland for a triaxial tester20 which will allow either an axial or torsional load to be applied to the specimen; Figure 14 shows application in a stirrer allowing rotation or axial motion. Shown in Figure 15 and Figure 16 are needleless, packingless throttle valves; these valves are self-cleaning, highly erosion resistant, and can be closed tightly, so a no-leak condition.
The controlled clearance principle is the same in all of these devices. The illustration in Figure 12 shows the controlled clearance used in a free piston gage. The controlled clearance is at (e) and is controlled by jacket pressure (Pj); (a) is the jacketing cylinder; (c) is the jacketed cylinder; (b) could be in principle the piston of the piston gage, the closure of a vessel, the shaft of a stirrer rod or the ram for axial or torsional loading of specimens in a triaxial test or the passive element of the controlled clearance device as shown in Figure 13, Figure 14, and Figure 15.
Figure 16 shows the valve configuration with the jacket pressure applied inside the controlled element, not outside as in Figure 15. This arrangement has occasionally some distinct operational advantages.
The vessel shown in Figure 17 was produced in the late 1950's. The vessel has a working cavity of 460 mm dia. x 1530 mm long; its external dimensions are 1260 mm dia. x 3090 mm long. The vessel has a working pressure of 3 ½ kilobar and was tested to 5 kilobar; the closure seals are a simple "O" ring backed by a wedge ring.
The vessel shown has a screwed closure retainer. Today's practice would be to recommend a quick opening device; we have made several designs.
One design is to put the vessel in a press frame, and using the fluid power equipment usually associated with the high pressure pumps to manipulate the vessel in and out of the press frame. Such an arrangement is shown in Figure 18. There are no load-carrying threads in this arrangement the end thrust is taken completely by the press frame.
The unit shown in Figure 18 operates at 7 kilobar and was pressure tested to about 8 ½ kilobar. Its working space is 254 mm dia. x 610 mm long; the outside dimensions are 1050 mm dia. x 660 mm long. The closure packings on this unit, too, are a conventional "O" ring with a wedge ring back up. Typically, two high pressure cylinders for compactions are provided; with proper materials handling equipment, while one cylinder is being pressurized, the second one is being unloaded and a new charge is being put into it. This design of equipment, in sharp contrast to the conventional screw retainer used on the older vessel, is adaptable to complete automation and high production rates.
The press frame is a typical Harwood "O" frame type. It is made of two steel plates with a central opening that is shaped carefully, and is calculated to minimize the stress concentrations to be expected. This particular frame has a rating of 5500 metric tons with essentially infinite life.
A third large vessel, using the same simple packings as above is shown in Figure 19. Here the longitudinal thrust is carried by the outer cylinder. The force is transmitted by a conventional threaded closure retainer at the bottom end where access to the bottom is seldom required. At the top a through pin transmits the load to the vessel. This shear pin is inserted and retracted by an hydraulic cylinder, powered by a portion of the fluid power drive circuit.
Among other useful control schemes is one to allow two single-acting intensifiers to deliver a pulse-free flow, even with gasses.22 This is particularly helpful in chemical processing as well as jet cutting.
One of the applications for high pressure pumping equipment is to supply pressures needed for the autofrettage of cylinders to be exposed to high pressure during their service.23 The autofrettage of these cylinders has significantly increased the fatigue life of the cylinders. The photograph shown in Figure 20 shows a new autofrettage system recently installed at the Japan Steel Works. The last stage intensifier itself is capable of reaching 20 kilobar but the system itself is limited to 15 kilobar, because of the limitations of the available tubing and fittings. The use of these intensifiers as the pressure generating equipment for fatigue studies is a natural extension of their use for autofrettage of weapons and industrial pressure vessels.
The major player in the studies of fatigue have so far been the Ordnance people; by necessity, they are interested in the mortal end of the fatigue curve. Guns are highly stressed by hot, eroding, chemically active gases at high pressure and consequently limited service life. Modern designs are moving toward higher pressures, higher strength, and lighter and more mobile weapons. Fatigue testing is an obvious necessity. Even so, it was clearly apparent form fatigue studies by the author, early in World War II, that autofrettaged cylinders had a fatigue life at least three times greater than nonautofrettaged cylinders. The Department of Defense has continued these fatigue studies. To this end Harwood has supplied fatigue equipment, to the the defense department, with displacements up to 3000 cc with 225 horsepower for pressures up to 7 kilobar. This large displacement allows the study of relatively large cylinders in fatigue.
Intensifiers for this service generally have a long life, despite the severity of their service; Harwood has some in the field working since the mid-1950's. Harwood installed a small 7 kilobar fatigue machine in Japan in 1975 which is used to qualify steels for service in high speed rollers. As of this time, it is still working fine. Another unit of interest is a miniature fatigue machine supplied to a government agency to cycle dynamic pressure gages at 10 cycles per minute to 10 kilobar. This was not a fatigue study in the traditional sense, but part of a program to define the expected life and accuracy of dynamic pressure gages.
As high pressure food technology reaches to process at more extreme pressures, the problem of fatigue especially with a potable water medium will become more and more important. As is well-known, water service greatly shortens the fatigue life.26 Modern work abounds with studies in the effect of stress corrosion.
Harwood has faced the problems caused by fluid behavior for 45 years. An early example of some of the difficulties: Harwood had a client who wished to do isostatic compactions at very modest pressures, but wanted to use standard hydraulic oil it has many good properties at room pressure that you would like in a pressure fluid it lubricates; it has reasonable viscosity; it is relatively inert, inexpensive; and, at ambient pressure at least, it transmits pressure very well. The system, however, would not deliver enough pressure. The pressure in the vessel, remote from the pump, would not increase beyond 35000 psi while the pump was at 50000, despite being connected with relatively large 5/16 inner dia.. tubing. However, the connection was long too long for the oil to transmit pressure readily. If kerosene, which transmits pressure easily to a higher level, had been used in the pump, this would not have occurred. An early researcher claimed reaching astronomical pressures (1 million psi) by looking at the ratio of areas in his intensifier and multiplying his drive pressure; Dr. Bridgman quietly reminded him that his particular high pressure fluid became solid at approximately 4 kilobar (60,000 psi).
At Harwood we are working on finding appropriate fluids for high pressure work. Among the properties we have been looking at:
A new tool for studies of fluid pressures is our step pressure generator24, capable of applying up to 10 kilobar to a sample in less than a millisecond. (See Figure 23). Originally developed for calibrating dynamic pressure gages, the device shows promise for testing the effect of high strain rates on viscosity, transmissibility, and also on the PVT properties of the fluids of interest. The speed of the application of pressure approaches what is needed for a true isentropic compression. High strain rates could well be important in biological phenomena. Strain rates were found to be important in the mortality of some bacteria in a few unpublished tests made at Harwood in the 1970's with Dr. Naomi McGovern of Sudbury, Massachusetts.
Harwood developed equipment25 for the measurement of viscosity to 30 kilobar, and have published some data. We found that in the pressure range from atmospheric pressure to 30 kilobar, the viscosity of white gasoline increased by seven orders of magnitude.
Still, for all of that, there is a great void in the understanding of the behavior of many materials physically, chemically, biologically at these lower pressure levels, of up to 30 kilobar. This should be an orchard easily and readily harvested.
