Airship Buoyancy Control Using Inflatable Vacuum
Transcription
Airship Buoyancy Control Using Inflatable Vacuum
JOURNAL OF AIRCRAFT Vol. , No. , Airship Buoyancy Control Using Inflatable Vacuum Chambers Sean A. Barton∗ Florida State University, Tallahassee, Florida 32306 DOI: 10.2514/1.C031654 1 Buoyancy control is identified as a significant problem in airships, past and present. An existing proposal to enhance the control of buoyancy in lighter-than-air aircraft is reviewed, and a new method of buoyancy control (the inflatable vacuum chamber) is proposed. The differences between the existing proposal and new proposal are highlighted, including the new possibility of an airship that lands vertically onto the ground without a mooring mast, a ground handling party, or a runway. Theory of lightweight stiff inflatable structures is discussed and a specific design and construction method is proposed. The design is optimized and confirmed elastically stable; design specifications are given. A typical operation cycle (including liftoff, flight, and touchdown) is detailed. An area of potential theoretical advancement is identified. I. techniques involving pressurization or compression appear to be more practical than those involving heating or liquefaction. The new Aeroscraft (Worldwide Aeros Corporation) is anticipated to use compressed air storage to control buoyancy reversibly. The Aeroscraft will be outfitted with high-pressure tanks in which air can be pumped to gradually increase the weight of the ship. When the tanks are full, weight can be discharged rapidly by venting the tanks. This capability could assist with fuel weight compensation, vertical ground takeoff (instead of mooring), and rapid loading of cargo. But without a method of rapidly adding weight the operations of vertical ground landing and rapid unloading of cargo remain difficult. These cannot be accomplished with a compressed air tank, because air pumps are too slow or too heavy. It can be shown that if the pump is of equal weight as the storage container, it could take hours to fill the tank, yet only seconds to empty it. Thus, rapidly and reversibly adding weight to an airship in-flight can be accomplished only with a tank of negative pressure. Unfortunately, conventional vacuum chambers are impractical for this purpose because of their large mass (high weight). However, in a recent development, the inflatable vacuum chamber (IVC) [2,3] allows for vacuum chambers of smaller mass (low weight) and, thus, it is now interesting to explore the potential application of vacuum to airship buoyancy control. The idea of using vacuum in an airship is not new. It dates back to 1670 when it was first proposed by Italian monk Francesco Lana de Terzi [4] (see Fig. 2). We now introduce this new possibility for buoyancy control, the IVC. Buoyancy control using an IVC is most closely related to the before-mentioned techniques involving pressurization or compression. These techniques will be briefly reviewed first. The IVC will then be discussed. Introduction E DWIN Mowforth [1], Vice President of the Airship Association, vice president of the Airship Heritage Trust, states, “Many of the airship’s principal operational problems are associated with the functions of buoyancy control : : : ” This fact is demonstrated in airship history. The U.S. Ship (USS) Akron was lost when it was forced into the sea by a strong downdraft. The USS Macon was lost after its captain, in a panic over a potential loss of buoyancy, ordered a massive and irreversible discharge of ballast; the Macon rose above pressure height (the altitude at which the gas bags expand beyond the total volume of the airframe), lost lifting gas, and eventually crashed into the sea. The loss of the USS Shenandoah might have been caused by airframe damage resulting from a breach of pressure height. Of the U.S. Navy’s airships, only the USS Los Angeles met its end of old age; however, it too suffered from an inability to adjust its weight to meet changing conditions (see Fig. 1). In modern airships, the usual buoyancy control problem is that of compensating for the weight of fuel consumed in-flight.† Inability to control buoyancy effectively also creates difficulty with landing, takeoff, and load exchange. Difficulties with the weight of fuel consumed during flight can be handled in many ways. Fuels of various densities (gases and liquids) can be used (as in the LZ 127 Graf Zeppelin). Vectored thrust and flight control can compensate to an extent. Difficulties with landing and takeoff are often avoided by not landing and instead mooring some tens of feet above the ground. An airship that actually lands and takes off can solve some of the buoyancy problem with aerodynamic lift, but this requires the airship to make a run along the ground (a runway) for takeoff and landing and, thus, erodes the airship’s prized land anywhere ability. Airships can also discharge ballast on liftoff, but discharge of lifting gas on touchdown is cost prohibitive. Load exchanges must be balanced, thus a load cannot be dropped except where sufficient ballast is available. There are a variety of standing proposals to allow an airship to reversibly change its weight in-flight. They include pressurization or heating of the entire gas contents of the airship, compressed storage of the lifting gas or air, and liquefaction of the lifting gas or air. A good review of these techniques can be found in [1]. It is concluded that II. Buoyancy Control Using Compressed Air Storage First, we explore the idea of storing compressed air on the airship. For such a system it is important to know the weight of the container and the weight of air it will be able to store. Often the limiting factor for a buoyancy control system is its weight. An ideal system would allow the airship to change its weight by a large amount quickly while adding only a small amount of weight to the airship itself. We first propose that the air be stored in long cylinders of continuously wound fiberglass composite. pInsuch cylinders, the glass fibers are wound at an angle of arctan 2 ≈ 54.74 deg to the Presented at the SDM 2011, Denver, CO, 4–7 April 2011; received 6 September 2011; revision received 5 January 2013; accepted for publication 21 January 2013; published online XX epubMonth XXXX. Copyright © 2013 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved. Copies of this paper may be made for personal or internal use, on condition that the copier pay the $10.00 per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923; include the code 1542-3868/YY and $10.00 in correspondence with the CCC. *Postdoctoral Associate, Department of Physics, 77 Chieftan Way. Member AIAA. † Personal correspondence with Edwin Mowforth, 14 August 2010. axis of the cylinder to provide strength around the circumference that is twice the strength along the axis. (When a cylinder is subject to hydrostatic loading the stresses in the circumference are twice those along the axis.) It can then be shown that the total weight of the container (excluding end effects) is σ composite −1 3P mtank mair-stored air (1) ρair ρcomposite 1 2 BARTON Fig. 1 The USS Los Angeles after being struck by a tailwind while moored on 25 August 1927. container is in tension only (no compressive forces). (For a proof of the previous statement, see the Appendix.) Because Eq. (1) is applicable to general shapes it is also valid for pressurizing the entire volume of lifting gas in the airship. In that case, however, one considers the stress-to-density ratio of the fabric (not composite) that carries the pressure load in the surface of the airship. Mowforth ([1] pp. 360–363), appears to contradict the idea that Eq. (1) is valid for pressurizing the entire envelope of the airship. He, however, acknowledges in private correspondence with the present author‡ that the choice ks 10 in ([1] p. 362), does not follow from any known discussion and is likely in error. It is the belief of the present author that Eq. (1) is applicable to any pressure container, including the entire envelope of a nonrigid airship.) For air at 15°C, Pair ∕ρair ≈ 0.083 km2 ∕s2 . For a 60/40 composite of E-glass/epoxy loaded at a safety factor of 4, σ composite ∕ ρcomposite ≈ 0.25 km2 ∕s2 . Thus, the mass of the tank will be approximately equal to the mass of the air the tank can store. If one substitutes a 70/30 composite of aramid/lightweight epoxy loaded at a safety factor of 4 we find σ composite ∕ρcomposite ≈ 0.47 km2 ∕s2 and, thus, the mass of the container would be about one-half the mass of air stored. It appears that containers of wound carbon fiber might have a mass 2.5 times smaller than the mass of air they can contain with the same safety factor. Energy consumption of the required air pumps will likely not exceed 3% of the airship’s total energy budget [1]. A survey of air compressor pumps (and vacuum pumps) on the market today indicates that they can easily pump a weight of air equal to their own weight in about 45 min. If engineered for minimal weight (for airship application) this could likely be improved by a factor of 2. III. Fig. 2 Francesco Lana de Terzi’s flying boat concept circa 1670. where m is mass, P is pressure, ρ is mass density, and σ composite is the strength of a unidirectional sample of the composite material on its strong axis. Note that Pair ∕ρair depends only on the temperature and not on the storage pressure (ideal gas law). It is also true (and more difficult to show) that Eq. (1) is also valid for containers of arbitrary shapes including cylinders with hemispherical ends, spheres, and ellipsoids as long as the material comprising the structure of the Inflatable-Vacuum-Chamber Concept We now shift our attention to the potentially complementary system, the inflatable vacuum chamber. We can begin to understand the mechanics of the IVC by considering the following idealized structure. One imagines that there is a dense three-dimensional isotropic network of thin tensile fibers. These fibers are placed randomly and densely, having all possible orientations and approaching a continuum. These fibers can be imagined to be glass, aramid, carbon, or otherwise. One then imagines removing regions of the tensile fibers to leave behind structures of a desired shape. For example, one might imagine leaving behind a spherical container consisting of the fiber segments in a volume between two concentric spheres of different radii and all other volumes void of fibers. One then imagines placing gas-impermeable barriers along every surface, where the volume of removed fibers meets the volume of remaining fibers. For the example of the spherical container mentioned previously this would consist of two spherical membranes coincident with the before-mentioned spherical surfaces. Because the fibers carry only tension and not compression the space between the fibers must be pressurized to provide the idealized solid with the ability to transmit compressive forces. One thus imagines that the severed ends of the fibers are mechanically connected to the membrane, so that the space occupied by the fibers can be pressurized, and the resulting tensions in the fibers will be terminated in the membrane (see Fig. 3). The distance between the fibers is considered to be small compared to the thickness of the membrane, so that the stresses in the membrane can be neglected. Note that the membrane does not carry any of the stresses resulting from pressurization. The force imparted to the membrane by fiber attachments is balanced by the pressure force. A structure such as this is filled mostly with gas and, thus, has a low density. Pressure and fiber cross-sectional areas can be reduced proportionally to reduce the effective strength of the structure. This reduction, however, does not affect the effective strength-to-stiffness ratio of the idealized continuous solid or structure that it forms. This is in contrast to traditional compressive structures which, when designed for lighter loads, lose stiffness at a proportion much greater than the proportion by which they lose strength. Corrugations and complex networks of compressive members are often used to address ‡ Personal correspondence with Edwin Mowforth, 13 March 2011. 3 BARTON We average this for all orientations of the fibers to find the average elastic energy per unit volume of fiber material: Z Z 1 π π u sin θ dϕ dθ 4π 0 −π 2 εx εy ε2y εy εz ε2z εz εx ε ≈E x 10 15 10 15 10 15 u Fig. 3 A cross-sectional view of the idealized spherical IVC: the thin lines represent tensile fibers and the thick lines represent gasimpermeable membranes. this problem, but as the design load further lightens, it becomes nearly impossible to maintain the ratio of stiffness to strength, hence the difficulty in designing lightweight vacuum containers as compressive structures. This is the motivation for the idealized inflatable solid just described. Its design load can be reduced arbitrarily while maintaining its strength-to-stiffness ratio. Thus, it can then be shown that, if this idealized spherical shell is sufficiently thick, the linear mass density of the fibers is sufficiently small, and the space between the fibers is sufficiently pressurized; the structure remains elastically stable when vacuum is introduced into the central space. Let us further consider this pressurized network of fibers. In the idealized case, a pressurized matrix of tension fibers can be described mechanically as a continuous solid having a specified strength and elasticity. We will here show that, if the tension fibers are isotropic, the idealized bulk solid has isotropic elasticity and Poisson's ratio of one-fourth. We will also show that the elastic modulus per mass and single-axis strength per mass of this idealized continuous material is one-sixth that of the fiber material. We imagine a dense isotropic network of thin elastic tensile fibers having all possible orientations: 2 3 cos θ a 4 sin θ cos ϕ 5 (2) sin θ sin ϕ We identify this as the elastic energy of an isotropic elastic solid with an elastic modulus of E∕6 times the fiber volume fraction and Poisson's ratio of one-fourth. This consideration is valid only if the strain on all three principal axes is positive. Under negative strain, some fibers relax to zero tension and the analysis become nonlinear. Notice that with such a network of fibers it is not possible to have zero tension along any single axis without having zero tension along all axes. This is because only a very small number of fibers carry tension along a specific axis or in specific planes, and none of the fibers carry compressive force. This means that such a tensile network always has a background tension, the transmission of which requires extra weight and extra pressurization to create force balance. For an example, consider the hydrostatically compressed spherical shell. The ratios of the tensions on the principal stress axes in the shell are 0∶1∶1 (radial∶circumferential∶circumferential). The idea of the pressurized tension network is to use pressures to achieve the isotropic positive part 1∶1∶1 and then use the tensile network to provide the additional reduction −1∶0∶0 needed to bring the total to 0∶1∶1. However, the requirement of only positive strain requires a “background tension” of 1/2 and, thus, we must substitute −1∶0∶0 with −3∕2∶ − 1∕2∶ − 1∕2. Thus, pressure (and the initial loading of all fibers) has to be increased by 50% to provide the background tension to prevent negative strain on loading. These difficulties can be reduced, however, if the fibers are not placed isotropically but instead are placed primarily along the anticipated axes of principal strain when loaded. Unfortunately, this approach increases difficulties with elastic stability and the analysis thereof. Because stress in the tensile fibers will be isotropic when the IVC is vented and will be primarily in the radial direction when vacuum is present we find that practical IVCs have a large population of tensile fibers in the almost radial direction. This almost direction is necessary to mitigate issues of elastic stability. The structures of Figs. 1 and 7a in [2,3] (reproduced as Figs. 4 and 5, respectively) illustrate this philosophy of exactly radial and almost radial, respectively. In this example, the tensile fibers are part of a textile membrane. We propose the structure of Fig. 5 for use in buoyancy control. Though spherical IVCs have better theoretical properties we avoid them for difficulties in automating their construction and difficulties in producing double-curved (spherically curved) fabric membranes. for all θ and ϕ. We then imagine stretching the network of fibers on three principal axes of strain resulting in new orientations: 3 2 0 0 1 εx 0 5a (3) 1 εy b4 0 0 0 1 εz where is the strain. ε These new orientations have elastic energy per unit volume of fiber material equal to 1 u Ejbj − 12 2 (4) where E is the elastic modulus of the fiber material. To second order in ε this is found to be 1 u ≈ Eεx cos2 θ εy sin2 θ cos2 ϕ εz sin2 θ sin2 ϕ2 2 (5) (6) Fig. 4 Upper right quadrant of Fig. 1 from [2,3]. 4 BARTON Fig. 7 Method of constructing the proposed structure. Fig. 5 IV. Figure 7a from [2,3]. Proposed Structure The structure has a cross section shown in Fig. 6 and is invariant in its length. In the figure, the lightly shaded exterior region represents low-pressure helium, heavily shaded regions represent high-pressure helium, the unshaded interior region represents vacuum, thick lines represent gas-impermeable fabric, and thin lines represent gaspermeable fabric. Its inner radius is about 1 m, and its total length (continuous or noncontinuous) is approximately 600 m. It provides about 7.5 tons of buoyancy control range. The ends are closed crudely by wetting the fabrics near the end of the structure with a sealant, placing a constricting band around the structure at that point, and then tightening the band until gas leakage is minimized. Ports and ducting are added to facilitate inflation and deflation. It should be noted that buoyancy control systems that occupy large volumes create different amounts of buoyancy at different altitudes even if their gas contents do not change. For example, because the proposed vacuum volume displaces about 2 tons of air at sea level and does not change its volume at altitude, its buoyancy is reduced by 0.62 tons at an altitude of 10,000 ft. V. Construction of the Proposed Structure The proposed structure is formed of five layers of fabric. The innermost and outermost fabric layers (layers 1 and 5, respectively), are gas impermeable, that is the fabric has been impregnated with some gas-impermeable resin or laminated with some gas-impermeable layer. These five fabrics are sewn together with a lock-stitch sewing machine as shown in Fig. 7. The structure is closed by bonding each of the five layers back to itself with a butt joint secured with an adhesive. The structure is preliminarily inflated at low pressure to give it shape. Lines of sealant are then placed between the lobes on the sewn seams to slow the gas leakage passing through the holes created by the sewing needle. The author has built preliminary structures of polyurethane-coated polyester fabric. The fabric layers were sewn together as described and sealed from the exterior with a paintable two-part polyurethane sealant. After three coats of polyurethane sealant were applied leak rates were found to be low. Fabric layers were closed onto themselves by heat welding of the polyurethane fabric coating. Special sewing machine fixtures were designed to aid in maintaining correct dimensions. A superior way of bonding two fabrics edge to edge is by means of a butt strap adhered to the face of each fabric with an adhesive. This is the current state of the art in the manufacture of nonrigid airships [5,6]. Although sewing the fabrics together is simpler it does not allow for a natural termination of the tensions in the fibers that are cut along the cut edge of the fabric. This can result in the position of the fibers shifting over time, especially in the fibers parallel to the seam that are between the stitching and the cut edge of the fabric. Shifting of the fibers is often slowed by the immobilizing effect of the sealant, but this results in the undesirable situation in which the sealant is transmitting part of the structural load. This usually results in failure of the sealant and leaks. In general, sewing together the cut fabric edges is undesirable for these reasons. In construction of the vacuum tank design proposed in this work there are five of these butt joints (one to close each of the five layers of fabric onto itself), and it is recommended that these butt joints be made in the discussed state-of-the-art way. The remaining joints in the structure (approximately 100 in number) are fundamentally different. They cannot be created with adhesives because the stresses present tend to peal the fabrics apart and fail the adhesive. Additionally, the known difficulties that follow from sewing of cut fabric edges do not apply because the fabric does not have cut edges at these joints. The present author recommends that these 88 structural joints be created by sewing the several layers of fabric together and applying a sealant over the stitches from the inside of layer 1 and the outside of layer 5. VI. Fig. 6 Cross section of the proposed structure. Stability of the Proposed Structure The stability of the proposed structure is analyzed with techniques similar to those found in [2,3] with the following exceptions. The effective-tension concept is not used. The angles of arc in the outer and inner lobes are added as additional degrees of freedom bringing the total number of degrees of freedom per unit cell from four up to six. The design strength of each membrane is calculated from the tensions present in the absence of vacuum. These tensions do not have a unique solution, until one imposes the additional constraint that the tensions in layer 2 and layer 4 must be equal. This constraint is arbitrary. It may be found that releasing this constraint will allow the optimized mass of the structure to be slightly lower and, thus, further investigation is required. The elasticity of each membrane is set by assuming that the elastic modulus divided by the tensile strength is 34 (characteristic of aramid fiber fabrics) [7]. The outside radius, number of lobes, and pressure are then varied keeping the inside radius fixed to find the structure with the smallest total membrane 5 BARTON Fig. 8 Failure mode of the proposed structure. mass per unit volume of vacuum that has both elastic stability and positive tension in all membranes. As shown in the Appendix the mass of gas the structure contains when vacuum is not present is directly proportional to the total membrane mass. Thus, the weight of the pressurizing gas is not considered when optimizing the structure. The optimized structure is found to have an outside radius that is 1.50 times its inner radius, 44 lobes, and a pressure of 3.08 absolute atmospheres in its wall. The failure mode of the structure is shown in Fig. 8. Tables 1 and 2 give the specification of the specific design. VII. System Operation The 2 t system proposed here (schematically shown in Fig. 9 for a nonrigid airship) is designed to aid with vertical takeoff (or liftoff), weight compensation for fuel consumed during flight, and vertical landing (or touchdown). Larger systems would be required to adjust buoyancy for onloading and offloading of cargo. The IVC functions as both a vacuum chamber and a pressure vessel. Its weight efficiency as a pressure vessel is equal to that of pressure vessels of simpler designs (see the Appendix). It, however, has the additional ability to contain vacuum (in addition to pressure) Table 1 when sufficient pressure is present. An IVC buoyancy control system thus has the same weight and buoyancy control range as a system that only compresses gases but has the additional benefit of being able to rapidly add weight. This comes at the cost of having a storage vessel of more complex geometry. We consider here an airship displacing approximately 100 t of air at sea level having a fuel capacity of about 5 (metric) tons. It experiences a drag force of about 2 t (metric ton force) at its top speed of about 100 km∕h. It is equipped with a 2000 hp engine and has a range of 1000 km. It flies at altitudes up to 10,000 ft. We begin with the airship on the ground, at sea level, no vacuum in the center of the IVC, full pressure (about 30.6 psi gauge) in the wall of the IVC, and no fuel. In this condition, the airship has been designed to have a net buoyancy of −2.00 tons, giving it traction with the ground. It is not moored. In preparation for liftoff 5.55 tons of fuel are added. The net buoyancy is now −7.55 tons. As time for liftoff approaches, engines are started and engine exhaust is directed through the heat exchangers warming the helium contents of the airship to produce 5.55 tons of superheat buoyancy. One notes that the superheat must be maintained only until the moment of liftoff at which time the pressure/vacuum system will add the buoyancy to the airship permanently. Thus, the difficulties of maintaining superheat for extended periods or at high velocity are avoided. Although it appears feasible to augment buoyancy with superheat for a short period of time [1], more investigation is needed to confirm this. The net buoyancy is now −2.00 tons. When the ship is prepared for liftoff the wall of the IVC is vented releasing compressed helium into the envelope of the ship. The helium cools adiabatically as it expands. The increase in lifting gas volume rapidly forces air off the ship. (In a nonrigid ship this air exits through the overpressure protection valves.) Buoyancy surges by 4.53 tons. Net buoyancy is now 2.53 tons. In absence of vertical winds or assistance from vectored thrust this produces about 1∕40g of vertical acceleration. Thus the airship rises to an altitude of 100 ft in about 15 s. Immediately after liftoff the heat exchangers are closed and thus heating of the lifting gas is discontinued. The airship's lifting gas now contains a pocket of hot helium (resulting from the lifting-gas heating) and a pocket of cold helium (resulting from the adiabatic expansion of the helium released from the IVC). As these temperature variations dissipate to the atmosphere during the next 15 min, 2.48 tons of buoyancy are lost, bringing the net buoyancy to 0.00 tons. The IVC system is now fully discharged. Simultaneously, as the airship climbs to 10,000 ft, the valves venting the interior and General specifications for the inflatable vacuum chamber Parameter Value General design specifications Radius to inside sewn seam 1.00 m Radius to outside sewn seam 1.50 m Number of lobes 44 Absolute pressure in wall 3.079 atmospheres Absolute pressure in central space As low as 0.00 atmospheres Safety factor 4 Total noncontinuous length 581.1 m Volume of central space 1632 m3 Pressurized volume 2965 m3 Total volume 4597 m3 Fabric of epoxy-impregnated aramid fibers Density of fibers 1.44 g∕cm3 Strength of fibers 3.6 GPa Elastic modulus of fibers 124 GPa Total fabric weight 2998 kg Total weight of axial fibers 999 kg 1∕3 total Total weight of perpendicular fibers 1999 kg 2∕3 total Fabric densities Layer 1 36.16 g∕m2 (not including gas-impermeable resin or layer) Layer 2 122.68 g∕m2 Layer 3 45.94 g∕m2 Layer 4 122.69 g∕m2 Layer 5 54.24 g∕m2 (not including gas-impermeable resin or layer) 6 BARTON Table 2 Weight components of an operational inflatable vacuum chamber Weight components Fabric weight Extra helium required to pressurize Helium removed when evacuated Total weight added to airship when fully charged Reduction in payload assuming discharge on takeoff Total buoyancy control range Weight of air displaced when central space is evacuated Weight of air undisplaced when central space is vented Weight of air immediately undisplaced on venting Weight of air displaced when wall is vented Weight of air immediately displaced when wall is vented Table 3 Initial state 2.998 tons 1.042 tons 0.276 tons 3.764 tons 2.998 tons 7.552 tons 2000 kg 2000 kg 1429 kg 7552 kg 4531 kg Buoyancy control capabilities of the inflatable vacuum chamber Immediate weight change Pressure Vacuum First Second C C 1.43 t 0 C C 1.43 t −4.53 t C C −3.10 t 0 C D −4.53 t 0 Delayed weight change 0.57 t −2.45 t −2.45 t −3.02 t Final state Pressure Vacuum C D D D D D D D C charged, C discharged. Note: Delayed weight change occurs as adiabatic heating/cooling dissipates. wall of the IVC remain open allowing the reduction in ambient pressure to further evacuate all parts of the IVC, thus buoyancy remains balanced. Once an altitude of 10,000 ft is reached, valves to the IVC are shut, locking-in the density of the IVC system. Because of the fixed volume of the IVC the airship will tend to stay at 10,000 ft. Avariation of 1000 ft in altitude will create a change in buoyancy of about 0.13 tons that will tend to return the airship to 10,000 ft. This oscillation of the airship's altitude will have an approximate period of 17 min. For the next 10 h or so, the airship flies its course of approximately 1000 km during which it consumes almost all of the 5.55 tons of fuel added. Simultaneously, a helium pump (thought to weigh about half a ton) removes helium from the interior of the IVC and from the envelope of the airship and forces it into the wall of the IVC to create a near vacuum in the interior of the IVC and a pressure of close to 28.0 psi gauge in its wall. This net compacting of the helium results in a reduction of buoyancy of close to 5.55 tons, precisely balancing the Fig. 9 Schematic representation of the IVC buoyancy control system for a nonrigid airship. weight of lost fuel. The net buoyancy remains at 0.00 tons, and the fuel tank is close to empty. As the helium pump continues to increase the gauge pressure in the wall of the IVC beyond close to 28.0 psi (where close to corresponds to the small weight of fuel remaining), the altitude for aerostatic equilibrium (which was 10,000 ft) begins to decrease, and the airship begins to descend. As the density and pressure of the ambient atmosphere increases, the buoyancy of the IVC system begins to rise and (in absence of continued pumping) the gauge pressure in the wall of the IVC would begin to drop. Thus, pumping continues to surpass changes in ambient pressure and bring the gauge pressure in the wall of the IVC to close to 30.6 psi as the airship approaches the ground. Net buoyancy remains at 0.00 tons. As touchdown approaches the pilot prepares to vent the vacuum in the central space of the IVC. When first contact with the ground is made an auxiliary air ballonet in the central space of the IVC is vented rapidly decreasing the net buoyancy by 1.43 tons. Net buoyancy is now −1.43 tons and, thus, traction with the ground is secured. During the next 15 min, as frictional heat dissipates to atmosphere, an addition 0.57 tons of weight are added to the ship increasing traction with the ground. For rapid return to the air, refueling and superheating of the lifting gas occur simultaneously. The auxiliary air ballonet is gradually emptied by allowing helium into the central compartment of the vacuum tank. Note that in this proposed system, the IVC vacuum space is finally filled with helium and not air. Note that the buoyancy of the ship is the same in either condition. In both cases, some gas already in the ship is moved into the IVC to make room for more air to flow into the airship. The addition of new air to the airship is what actually changes the weight of the ship. Additionally, moving helium instead of air into the IVC has the advantage of increasing the pressure height of the ship whenever vacuum is vented. This might be useful in the event that the IVC system is used in-flight for disaster avoidance. A similar argument shows the advantage of pressurizing the wall of the IVC with helium. Furthermore it is useful for the central space of the IVC and the wall of the IVC to contain the same gas, so that only one pump is required to manipulate the gas. Alternately, placing the vacuum system into an air atmosphere (and filling it with air) has the advantage of being able to rapidly exhaust gas to the exterior of the ship (which reduces the retention of thermally disturbed gas on takeoff, eliminates the need for overpressurization protection in a nonrigid airship, and simplifies structural considerations in a rigid 2 airship (see Sec. IX). It has the disadvantage of permanently reducing pressure height and/or payload capacity of the airship. VIII. Use of Oversized Vacuum Tanks Once the airship designer has specified the buoyancy change capability (2 tons in our examples) there still remains some choice in the total volume of the vacuum tank. Until now we have only considered tanks from which the vast majority of gas in the central space is removed. As an alternative, one can consider a tank of larger volume with less structural strength, so that only a fraction of the gas in the central space is removed. In any case, the same amount of gas is moved, the same change in buoyancy is achieved, and the structural weight of each tank is the same. The major advantage of the larger tank is a reduction in pumping energy because the same amount of gas is moved with less back pressure on the pump. This makes possible a faster or lighter pump and further reduces the already small amount of engine power required to recharge the system. As an example, we consider a 100 t airship having a 2 t vacuum buoyancy system inside the helium lifting-gas volume. If the vacuum tank is of minimum size (designed for full evacuation) the energies required to charge the pressurized wall and central vacuum are 492 and 276 MJ, respectively. (These energies assume fast adiabatic compression in the pump and slow isothermal compression in the tank. Actual energy consumed will be almost double because of mechanical friction in the pump.) An example alternative is to increase the total volume of the tank by a factor of 2 and remove only half of the gas in the central space. The energies required to charge the pressure and vacuum in this case are 7 BARTON 276 and 55 MJ, respectively. At four times the volume and one-fourth evacuation these become 148 and 24 MJ, respectively. Assuming adiabatic compression in the pump and isothermal compression in the tank, the energy E required to pressurize or vacuumize a volume V to a pressure P in an atmosphere of pressure P0 and inverse relative heat capacity at constant pressure Γ (2/7 for ideal diatomic gasses) is given in Eqs. (7) and (8), respectively: 1Γ P V PV P0 V P Epressure 0 − (7) 1Γ Γ Γ1 Γ P0 Evacuum P∕P0 − P∕P0 1−Γ 1 − P∕P0 1−Γ Γ1 − Γ (8) The benefit of minimizing energy required to charge the vacuum system is mainly in the weight and speed of the pump. A 2000 hp engine delivers 1000 MJ of mechanical energy every 11 min. Thus, in the extreme case when the flight is short (perhaps 4 h) the vacuum tank is small (requiring full evacuation), and the pump is 50% efficient; the energy required to charge the vacuum system is only 7% of the total engine power. Although compact vacuum systems require mechanical vacuum pumps and mechanical air compressors to reach the required pressures, systems of larger volume use roots blowers and multistage centrifugal blowers with attendant reductions in weight or increases in speed. A secondary benefit of using an oversized tank is the reduction in weight of the primary vent and primary exhaust valves (the large valves used at touchdown and takeoff, respectively, to provide rapid movement of gas and rapid buoyancy change) because of the reduced pressure differences. The rates of buoyancy change are limited by these valves (and the capacity of the overpressure protection louvers). Assuming a set amount of weight budgeted to these valves, vent times can be reduced by a factor roughly equal to the factor of increase in tank size, see Sec. XI. Pumping required to reach aerostatic equilibrium when altitude is changed is relatively unaffected by tank size. For example, when the ship moves from pressure height (11,865 ft) to sea level, ambient air density changes by 30% of sea level air density. When the vacuum system is in the lifting gas and compact (4597 m3 ) this corresponds to an increase in buoyancy of 1.689 tons, thus 1379 m3 of helium must be compacted away to maintain neutral buoyancy. This is about 22% of the total volume required to charge the pressurized wall. For a tank having twice the volume buoyancy change is doubled and, thus, an amount of gas equal to 44% of the total charge must be pumped, but this pumping occurs at roughly half the back pressure and with a pump of the same weight is thought to occur in approximately the same time. It can be shown [see Sec. XI, Eq. (14)] that, even though a tank of double the minimum volume has less pressure difference available to drive gas into the tank, its total fill time (with a valve of equal aperture) is less by a few percent. The natural oscillations of the airship's altitude when the vacuum system is locked in increase in frequency when larger tanks are used. IX. is difficult to provide a sufficiently large conduit from the rapidly deflating gas cells to the outside, so that the rapid movement of air does not put excessive compressive forces on the structure of the airship. A third option, which resolves both of these issues (and is the recommendation of the author), is to place inside the central space of the vacuum tank and auxiliary air ballonet. When the space is under vacuum, the auxiliary ballonet is empty. When the airship lands a primary valve is opened connecting the auxiliary ballonet to the outside atmosphere. The forces of vacuum rapidly fill the auxiliary ballonet with air creating the weight gain needed to secure the landing. A secondary valve, which connects the central vacuum space (outside the auxiliary ballonet) to the lifting-gas volume is then opened allowing lifting gas to gradually move into the central space of the vacuum tank thus slowly deflating the auxiliary ballonet. In a nonrigid airship this happens naturally under the slight pressure generated by the primary air ballonet blower and corresponds to an increase in the volume of the primary air ballonet. In a rigid airship, a small blower is used to transport the contents of one or several gas cells into the central space of the vacuum tank, thus making more volume for other gas cells to expand when altitude increases. This exhausting of the auxiliary ballonet must be completed prior to the airship returning to pressure height to provide enough volume for the altitude expansion of the lifting gas. When the ship again takes to the air, the helium in the wall of the vacuum tank must be released to create the surge in buoyancy that will provide a safe and rapid takeoff. The resulting increase in liftinggas volume forces air out of the ship. This change in volume is again too rapid for the air ballonet blower backflow (in a nonrigid ship) to discharge without overpressurization of the ship. In a rigid ship, it is also difficult to have such a large conduit from the rapidly inflating gas cells to the outside atmosphere as to prevent excessive tensile forces on the structure of the ship. Unlike the central space of the vacuum tank, placing auxiliary ballonets in the wall of the vacuum tank is not feasible, because the volume is divided into many smaller volumes by structural gas-permeable fabrics. If auxiliary ballonets were possible, one could pressurize the wall of the vacuum tank with the right proportion of air and helium, so that upon depressurizing the wall on takeoff only air is discharged from the wall, and only helium remains in the wall. In this way the discharge could be dumped directly to atmosphere and the pressure height of the ship would be preserved. Perhaps an inflatable vacuum tank of a different geometry can be designed, so that the pressurized volume is more consolidated; this is the subject of future research. For the present, the author recommends that the airship be outfitted with an overpressurization protection system. In a nonrigid airship, this might consist of a number of pressure relief valves (having a total exhaust aperture area roughly 10 times larger than that of the primary exhaust valve for an airship pressurized at 0.01 atm) connecting the primary air ballonet to the atmosphere, so that the ballonet can be exhausted sufficiently fast to ensure a safe and rapid takeoff. The exhausting of the vacuum tank wall could not exceed the volume flow capacity of the primary ballonet exhaust system. In a rigid airship, one would need to carefully consider which gas cells the wall of the vacuum tank should be exhausted into, and if the air around these rapidly expanding gas cells can escape the airship with sufficient ease to prevent structural damage. Management of High-Volume Gas Flows When the airship lands the central space of the vacuum tank must be vented to gain traction with the Earth. There are several ways this can be accomplished. A first option is to vent directly with atmospheric air resulting in direct weight gain. This first option has the disadvantage of reducing payload capacity or pressure height of the airship. The airship can reach full pressure height only if 100% of its gas volume is occupied by helium. The presence of any air volumes that cannot be exhausted results in a loss of payload capacity or pressure height. A second option (which resolves this first issue) is to vent with helium from the lifting-gas volume causing the volume of the lifting gas to decrease and, thereby, causing the air ballonet to increase in volume creating the increase in weight indirectly. With this second option, one finds that the air ballonet blower is generally unable to keep up with the massive volume of air that must be moved to prevent implosion of a nonrigid airship. Even in a rigid airship, it X. Isentropic Expansion vs Isenthalpic Expansion When calculating the buoyancy changes resulting from a release of vacuum or a release of pressurized gas the distinction between isentropic expansion and isenthalpic expansion must be considered. This was not considered in previous work [8]. When an ideal gas is expanded in a manner in which the maximum amount of work is extracted (e.g., with a piston) the temperature of the gas decreases, and the entropy of the gas remains unchanged. This is isentropic expansion. Similarly, isentropic compression of an ideal gas results in heating. Alternately, when an ideal gas is expanded in a manner in which no work is extracted (e.g., through a porous plug), the temperature and enthalpy of the gas immediately prior to and immediately after the expansion are the same (assuming that velocity and bulk kinetic energy is negligible). This is isenthalpic expansion. 8 BARTON As an example, we consider an evacuated vacuum tank being opened to the atmosphere (not inside an airship). In this generic case of a vacuum being released the expansion of the ambient atmosphere into the vacuum tank can occur either isentropically or isenthalpically. For the most rapid expansion, an orifice or nozzle is opened between the atmosphere and the interior of the vacuum tank. The work energy from the isentropic expansion of the gas accelerates the gas into the tank as it passes through the nozzle. Thus, the temperature of the gas decreases, but its velocity increases. The total of the thermal energy, kinetic energy, and potential energy (from atmospheric work) must remain unchanged as required by energy conservation. As this kinetic energy dissipates through the action of viscosity it is converted back to thermal energy and, thus, the final state is the same as if the expansion had been isenthalpic, wherein the kinetic energies are dissipated as they are produced inside the porous plug. In either case, the work that the atmosphere does to push the gas through the orifice, nozzle, or porous plug is finally found in the vacuum tank as thermal energy. As more gas enters the tank, gas already in the tank is compressed isentropically causing its temperature to rise. In the final state, when pressures inside and outside the tank are equal, we find that the potential energy of the atmosphere has decreased by an amount P0 ΔV atm , where P0 is the atmospheric pressure, and ΔV atm is the volume of the atmosphere, which has moved into the vacuum tank. Ideal gas properties for the gas in the tank in the final state give P0 V tank NkT tank Utank 5 NkT tank 2 (9) P0 ΔV atm NkT 0 (11) 5 U0 NkT 0 2 (12) In the final state, its internal energy must increase by the amount of atmospheric work: (13) Combining Eqs. (9–13) we find 7 T tank T 0 5 Dynamics of Filling the Vacuum Tank P v2 P 0 Γρ0 Γρ 2 where Γ≡ NkB CP 2 5 2 7 and thus we find s Γ 2P0 P v 1− Γρ0 P0 The rate at which volume is removed from the atmosphere is Avρ V_ ρ0 (14) where A is the area of the orifice through which the tank is vented. The average thermal energy in the tank during and immediately after venting is equal to the sum of the thermal energy and work energy that has entered the tank during venting (see Sec. X): 1 1 − 1 NkT tank − 1 NkT 0 P0 V Γ Γ 1 T0 NkT 0 → T tank Γ 1−Γ This heating of the gas (which comes from the work energy that was required to create the vacuum initially) results in the tank filling to only 3/5 and 5/7 of its total mass capacity for monatomic and diatomic gases, respectively. The tank fills to full capacity only once this frictional heat has dissipated. In a medium-sized airship this will take about 15 min. The pressure in the tank is given by 5 ΔV atm V tank 7 Thus, the change in buoyancy of the vacuum tank upon venting is only five-sevenths of its change in buoyancy on evacuation. The remaining two-sevenths of the buoyancy change is realized when the frictional heat has dissipated to the environment. In an atmosphere of helium (a monatomic gas), these fractions are three-fifths and two-fifths, respectively. Thus, for an airship with an inflatable vacuum tank inside its lifting-gas volume that is vented with air at the moment of touchdown 71% of the buoyancy decrease comes immediately; the remaining 29% comes gradually over the next 15 to 20 min (the time required for lifting-gas heat to dissipate in a medium sized airship). (The heat capacity of the fabric forming the vacuum tank is approximately equal to the heat capacity of the helium it contains when unpressurized, thus half of this heat might be soaked up by solids that the helium comes in contact with during the first 5 min.) The average temperature in the vacuum tank during the entire filling process is seven-fifths ambient temperature. Early during the filling, all of the gas is this temperature. Late in the filling, the gas that entered earlier is hotter, and the gas that entered later is cooler, but the average remains unchanged. for ideal monatomic gases for ideal diatomic gases Here, P and ρ are the pressure and density in the flow, respectively, and P0 and ρ0 are the stagnation/ambient pressure and density. The adiabatic condition requires 1−Γ P ρ ρ0 P0 (10) This same gas in the initial state also obeys ideal gas properties: Utank U0 P0 ΔV atm XI. When the vacuum tank is vented some time is required for the gas to move into the central space. Once the valve is opened gas accelerates toward the valve opening expanding adiabatically according to Bernoulli's equation for compressible fluids: Ptank P0 V T tank P1 V0 T0 where V is the total volume removed from the atmosphere, V 0 is the volume of the tank, and P1 is the pressure already present in the tank at ambient temperature (nonzero of oversized tanks). The time required to fill the tank is then V 0 1−Γ 1−P1 ∕P0 Z t 0 1 V 1−Γ dV 0 p A V_ Γ ss Γ 2ρ0 P1 1− P0 P0 Thus, when the tank volume is 1632 m3 and fully evacuated (not oversized) and fitted with a valve having an orifice area of 0.25 m2 (a valve weighing approximately 200 kg), the fill time for helium and air is 11.3 and 42.9 s, respectively. (Note that fill time for oversized tanks with a valve of the same weight is less.) Helium and air will fill the tank to three-fifths and five-sevenths full mass capacity, the mass defect being related to the frictional heat of filling the tank. When the medium is air, one might consider exhausting this heated air to atmosphere to replace it with air of ambient temperature to acquire the 9 BARTON remaining 2/7 of the capacity, however, one finds that to exchange this volume of air in 1 min requires a blower weighing approximately 1000 kg and a valve orifice of approximately 2 m2. Avalve of this size would weight approximately 1500 kg. Thus, it is concluded that the vacuum tank should be sized to give the desired weight change during its initial fast filling period when it is driven by the large pressure difference built up over hours by the work of the vacuum pump. XII. Failure of the Inflatable Vacuum System The vacuum tank can fail in two primary ways: leakage and explosion. First, the vacuum tank’s gas impermeable skin can fail, leaking helium into the lifting-gas volume of the airship. In the case of slow leaks, the helium pump can replace the leaking helium with no effect on the airship. In the case of faster leaks, the vacuum volume (if charged) can be used to temporarily take up some of the leaking helium thus temporarily preventing the buoyancy of the ship from changing. As pressure continues to drop in the wall of the vacuum tank the vacuum must be released to prevent geometric instability of the vacuum tank structure. Finally, when pumping and venting of vacuum can no longer keep pace with the leak, the extra helium will begin to accumulate in the lifting-gas volume forcing air off the ship. This loss of air will result in an uncontrollable increase in buoyancy. A discharge of lifting gas may then be required to prevent a breach of pressure height and safely land the ship. If the leak is fast or results from a violent tearing of the vacuum tank's fabric, the available pathways for exhausting air may be insufficient to prevent overpressurization of the airship and the resulting structural damage. In anticipation of such difficulties, one could potentially design the airship to fail in a very specific way when overpressurized, so that pressure is quickly relieved, damage is confined to a single area, and the remainder of the lifting gas is retained. This is the subject of future research. After the vacuum tank system has failed it can no longer provide the buoyancy control function required to land the ship. An emergency landing will be secured by other means (loss of aerodynamic lift [runway landing], mooring, ground handling party, vectored thrust, jettison of lifting gas, etc.). XIII. Future Theoretical Work When pressure in the wall of an IVC containing vacuum is sufficiently reduced instability will occur. This instability may potentially collapse a large fraction of the volume occupied by the vacuum. It might also be true that, immediately after the collapse, slightly increasing the pressure in the wall might restore the volume of vacuum. If such a tipping point phenomenon occurs it might be useful in airship design. If one can manipulate such a tipping point with only small changes in pressure, then one might move large volumes of gas with only small amounts of pumping. This may lead to a buoyancy control system that does not need charging, one that can switch back and forth from its heavy state to its light state at will by simply collapsing and uncollapsing its IVC at this tipping point of stability/instability. XIV. Conclusions In the current work, we have re-established the need for buoyancy control in airships, discussed existing proposals to control buoyancy with the storage of compressed air, re-introduced the inflatablevacuum-chamber (IVC) concept, optimized a specific IVC design for buoyancy-control application, considered methods of constructing the proposed structure, analyzed the elastic stability of the proposed structure, given specifications for the proposed structure, discussed the operation of the proposed buoyancy-control system, considered vacuum tanks of different volumes, discussed the logistics of venting the vacuum tank, discussed the heating and cooling effect that occur upon venting, discussed the dynamics of venting, considered the potential failure of the system, and illuminated potential avenues for future investigation. We continue to develop and test the techniques necessary to construct a working IVC approximately 50 m3 in total volume. We are also investigating experimentally the isenthalpic heating upon venting to gain a better understanding of the time scales involved in delayed buoyancy change. Appendix: Proof of Equation 1 We here prove that Eq. (1) is applicable to arbitrary shapes. We consider a number of tensile fibers bonded together with a resin to form tensile shells (or membranes). These tensile shells form structures, which contain pressurized gas. We assume that fibers are only placed where needed for strength and, thus, all fibers are at their maximum working load when the pressure is introduced. We first integrate the trace of the stress tensor σ ij over a volume that contains the entire structure (including all fibers, resin, and pressurized gas). We offset the stress tensor everywhere by the ambient pressure such that the pressures considered are gauge pressures, and the stress tensor outside the structure is zero: 3 2 0 0 Pgas Z 7 6 0 7 Pgas trσ ij dV tr6 5V gas 4 0 allvolume 0 2 Z composite volume 6 tr6 4 0 Pgas 3 −σ composite working load 0 0 0 7 07 5 dV 0 0 0 0 3Pgas V gas − σ composite working load V composite (A1) Note that the tension in the composite is shown to occur only in the x direction. This is clearly not true but also does not affect the result, because the trace is invariant under rotations of the coordinate system. Because the stress tensor at the boundary of the volume is everywhere zero, and because the divergence of the stress tensor is zero (momentum conservation), it is true that ∫ all volume σ ij dV 0ij . Also, because the trace of the integral is the integral of the trace, it is true that Z Z trσ ij dV tr σ ij dV tr0ij 0 (A2) all volume all volume and, thus, from Eq. (A1) V composite mgas 3Pgas V gas thus mcomposite σ composite working load 3Pgas ρcomposite ρgas σ composite working load (A3) which is Eq. (1). Acknowledgments The author would like to acknowledge Jack Sams for bringing together the right people to generate this idea, Ian Winger for developing techniques for constructing inflatable structure from fabrics, Helena Safron for editing this document and illuminating errors in the author's logic, Edwin Mowforth for sharing his lifetime of experience with lighter-than-air aircraft, and Frank Flaherty and Robin Winton for useful discussion. The author would also like to acknowledge David Van Winkle for making possible the trip to Denver to present this work, Juan Ordonez for administering the continued fabrication research, and the anonymous reviewer that has helped bring the manuscript into its current form. Reference [1] Mowforth, E., “Improvements,” Airship Technology, edited by Khoury, G. A., and Gillett, J. D., Cambridge Aerospace Series 10, Cambridge Univ. Press, New York, 1999, pp. 359–384. 3 10 BARTON [2] Barton, S. A., “Stability Analysis of an Inflatable Vacuum Chamber,” Journal of Applied Mechanics, Vol. 75, No. 4, 2008, p. 041010. doi:10.1115/1.2912742 [3] Barton, S. A., “Stability Analysis of an Inflatable Vacuum Chamber,” Journal of Applied Mechanics, Vol. 76, No. 1, 2008, p 017001. [4] Terzi, F., Prodromo: Overo, Saggio di alcune inventioni nuove premesso all'Arte maestra, Rizzardi, Brescia, Italy, 1670. [5] Bradley, P., “Materials,” Airship Technology, edited by Khoury, G. A., and Gillett, J. D., Cambridge Aerospace Series 10, Cambridge Univ. Press, New York, 1999, p. 151. [6] Zhai, H., and Euler, A., “Material Challenges for Lighter-Than-Air Systems in High Altitude Applications,” AIAA 5th Aviation, Technology, Integration, and Operations Conference, Arlington, VA, AIAA 2005-7488, 26–28 Sept. 2005. [7] Kevlar Aramid Fiber Technical Guide, DuPont, Publication H-77848 4/ 00, E. I. du Pont de Nemours and Company, Wilmington, DE, April 2000, http://www2.dupont.com/Kevlar/en_US/assets/downloads/ KEVLAR_Technical_Guide.pdf [retrieved 21 March 2011]. [8] Barton, S. A., “Airship Buoyancy Control using Inflatable Vacuum Chambers,” Proceedings from the 52nd AIAA/ASME/ASCE/AHS/ ASC Structures, Structural Dynamics and Materials Conference, April 2011. BARTON 11 Queries 1. AU: Please review the revised proof carefully to ensure your corrections have been inserted properly and to your satisfaction. 2. AU: As per journal style, section citations in the text denoted by their respective label numbers only. 3. AU: Please check and confirm the references and their respective citations in the text.