Tooling and Arbors @ Test Devices, Inc.
|
Tooling & ArborsIntroductionTest Devices has been designing and manufacturing spin tooling for over 30 years. In that time, spin tooling has been designed for a wide range of test rotors from industries including aerospace, power generation, automotive, and air handling, among others. Arbor design requirements which do not vary substantially from industry to industry and, like Rodney Dangerfield “don’t get any respect”, do entail advanced knowledge, different approaches and precise and detailed application to be safe and effective. Most of us think about tooling as a hunk of metal that is machined quickly and inexpensively to hold or support more critical parts. Because it is not a production part, delivered to a customer, it can be made of just about anything and loosely toleranced. Quite the opposite is the case with spin test tooling. As you may have read on other pages on our site, the amount of energy stored in a high speed rotating device is huge and can cause great damage. The tooling (arbors & spindles) is subject to the same speed induced stresses and therefore must meet at least the same performance specification as the customer’s part being tested. But that is not the whole story. The Heisenberg principle in physics basically says that anything that is watched is affected by the act of watching it, such as measuring one atomic particle with another atomic particle. In the same manner we must make sure our tooling does not significantly enter into and affect the test results. The act of holding a DUT (Device Under Test) will affect the internal stresses of the part. Test Devices’ job in creating tooling for spin testing is to mimic, as closely as possible, the manner in which the part is held and the stresses it is subjected to in actual operating conditions. This ensures that the test results are in line with real world conditions. Our job is to provide you, the customer, with data for real world product improvement. Essentially, spin tooling is a precision instrument and not just an afterthought hunk of metal. CONSIDERATIONSSome of the issues that we must consider in the design before starting fabrication include: Temperature: The effect of temperature has a profound effect on the material used and plays a significant role in the cost. Temperature consideration can be broken down in to four major categories: Ambient, Elevated-Temp (400° to 800°F), High-Temp (>800°F) and Cold (< 30°F). Each of these categories will determine which material to use. For example, Ambient temperatures can use certain carbon steels which are the least costly. These steels are the easiest to machine but may require a surface treatment to prevent premature failure from the effects of corrosion (ambient moisture, finger prints, etc). All materials are weaker as temperatures rise and therefore tooling for Elevated-Temp tests require higher strength materials, such as Stainless Steel. These Elevated-Temp tooling materials are more difficult to machine and usually require heat treating, which adds to the cost. Obviously, spin tooling is subjected to the most difficult conditions at temperatures above 800° F, which we classify as High-Temp. Again, because materials are weakened by increased temperature, we must specify certain materials for High-Temp applications to offset this loss in strength. Super alloys, such as Inconels, etc. are required for extreme elevated temperature tests. Not only are super alloy raw materials very expensive, but it is much more difficult to machine, and often requires special ceramic machining bits, as well as costly heat treatment procedures. Additionally, super alloys are only available in limited sizes which adds time to machine the excess material, again escalating costs. At the other end of the scale, very cold (cryogenic) tests also require precise material selection. Cold spin tests require the use of materials with transition temperatures lower than the test temperature. Parts can grow as much as one eight inch (1/8”) in diameter during testing and the loss in ductility below the transition point can cause cracks to grow quickly and result in a part (or tool) burst during testing. Each material has different cryogenic properties, and some steels, for example, become embrittled and thus dangerous to use at 0°F! Incorrect material selection, due to a lack of experience or understanding of the stresses and thermal effects, risks not only the success of the test but also the DUT.
Figure 1 Grip: Now that we have selected a material which will perform properly in our high stress environment, the next concern is how it will grip or hold the part. In designing tooling we try to understand the manner in which the test piece is held in actual operation. We can clamp it sandwich style (figure 2) or hold it around a feature (figure 3), either internally or externally. With the sandwich style we can loose the grip on the part should the preload not be sufficient. For this reason, sandwich style gripping is less suitable for Low Cycle Fatigue (LCF) tests and used mostly when replicating actual mounting conditions. As the part increases in speed, the diameter grows resulting in the shrinking of the overall length (Poisson’s contraction) of the part. This results in a loss of preload and loosening of the nut. Should the preload drop below a certain level during the test (different for each combination and is part of our upfront analysis), the vibration amplitude will increase and the test will fail. When analysis indicates that the preload could approach this level, TDI may install a preload enhancer (e.g. Belleville washer) to assure proper levels of preload. The downside is added complexity, less rigidity in the rotating assembly, and higher vibration levels.
Figure 2 & 3 Another method, and the one preferred by Test Devices, is to shrink fit the arbor onto the test part, taking advantage of the coefficients of thermal expansion. By cooling the DUT, and heating the arbor (or visa versa depending on the assembly) with precision and care in order to not affect material properties, we can create a clearance fit to facilitate assembly. As the arbor and DUT move toward equilibrium (come to the same ambient temperature) the assembly clearance gap is consumed, producing an “interference” fit. We have taken this tried and true technology (used for years in the railroad industry to attach wheels to axles) and applied additional science to it for more precision. This style of attachment has several advantages: It is very ridged, has lots of grip surface area for even distribution of stresses, and is very precise so that exact stress loading can be designed and predicted. In addition, the large surface and distributed stress provides a very powerful hold on the DUT. This style of joint also permits suspending the part (from the turbine) without additional hardware which reduces complexity and vibration. A thorough understanding of the coefficient of expansion of the materials selected is critical. If the interference fit is too aggressive, the arbor/DUT assembly may fail due to overstress. Conversely, using too loose a fit may result in the loss of interference during testing. Unlike the sandwich fit, with the proper selection of arbor material, we can assure that the tool and DUT grow at the same rate during testing, and avoid difficulties with the shrink fit joint. By using shrink fits, we can very closely replicate operating conditions with the right materials and tolerances. An alternative, though less attractive method of gripping, is to use the same approach as in figure 3, but to have only a slip fit and use a cross pin to hold all of the pieces together. Although this is a valid approach which is less expensive, it has limitations. The assembly will not be rigid and dynamic stability calculations will be more difficult and costly due to the need for iteration. The cross pin will take on a reverse load on each spin-up – spin-down cycle creating slippage and galling, making it unsuitable for LCF. It tends to produce a higher vibration level during testing due to the looseness of the joint, and its lack of rigidity will cause false positives when used with our Patented Crack Detection system during LCF tests. Many customer parts have a bolt flange where the arbor should be attached. Different from the pervious methods described, it involves a series of nuts and bolts arranged in a bolt circle and employs aspects of methods previously described. One of the most difficult issues with the bolt circles is assuring that the axis of the arbor is the same as the axis of the DUT (concentric). To enhance this alignment, we generally create an alignment boss on the arbor that will slip fit or interference fit into the hub of the DUT. Typically, the interference fit is not aggressive and does not carry the weight of the DUT and would not represent the stresses seen in operation. It is important that we define other aspects such as the torque, lubrication and preloads of each of the nuts and bolts to assure stresses are equalized and adequate. One must also consider the diameter of the bolt circle and the shear capability of the bolts.
Figure 4 In all methods, tolerances play an important role in performance and cost. A loosely toleranced part will cause the arbor dimensions to be more customized so that we can assure axis alignment and proper gripping. In these cases it also slows the process as tooling finalization must wait for delivery of the part for testing.
Figure 5 Test Devices has developed some special processes for sacrificial coatings. These are more often used on production tooling where the interference fits can alter through wear. With the use of the coatings, we can maintain the precision without the expense of new replacement tooling. Dynamic Stability - Critical Speeds: an absolutely perfect cylinder will spin without difficulty. In the real world, nothing is perfect and the slightest imperfections get greatly magnified at high RPMs. The effect of these imperfections is that certain sections of the cylinder will bow or otherwise deform. This deformation causes unbalance and can prevent the DUT from attaining its designed speed. Typically, the critical speed is related to the mass / stiffness distribution and we are able to accommodate for this in the design of the spin tooling by using software which analyzes the critical speeds of rotating assemblies. We spoke earlier about rigidity. Any movement of individual pieces under test (i.e. between the DUT and its arbor) can cause unbalance that makes calculating the critical speed extremely difficult and is one of many reasons why shrink fits are preferable. Vibration, in and of itself, is not the limitation, but vibration in resonance can destroy the test, DUT and equipment. Dynamic Stability – Moment Ratio: Why does a spinning top stay in one place and a spinning bowling ball tend to “walk” from its original position. There is a ratio that we calculate between the polar inertia and the diametrical inertia of a rotating assembly. Parts with an IP/ID ratio (polar to diametrical inertia) in a range near 1.0 are quite unstable due to the uncertainty as to which axis should predominate. The bowling ball fits perfectly into this band with IP/ID ratio of 1.0. Due to minor fluctuations, one axis may predominate over another and this is when the bowling ball “walks”. On the other hand, long cylinders (strong diametrical inertia) or flat plates (strong polar inertia) are very stable. Again, we are able to calculate this and design our tooling around this instability by varying the shape or mass of our arbor tooling. This is typically an iterative process between the critical speed calculations and the moment ratio, and is accomplished by modeling both the DUT and tooling in a solid CAD model and adjusting its mass properties as a system. Damping: Sometimes it is not possible to design around a particular problem due to configuration, environment or other factors. When confronted with an unresolvable prediction, we will try damping the tool assembly to reduce the amount of resonance. Assembly: In figure one, one notices that the nut has to be tightened onto the test assembly and we described a substantial pre-load. Another consideration during the design of a preload assembly is how to hold the assembly during torquing with out damaging the arbor or part. Depending on the actual configuration and magnitude of the torque, we can include features in the arbor design for holding or create more complex Torque Reaction Tooling when large torques are required. Even simple features, such as flats, must be precisely placed and closely toleranced to avoid causing vibration during the spin test. Materials: All spin tooling materials, whether exotic or not, must be specifically sourced, and cannot be supplied by all vendors. As discussed earlier, the forces and stresses become so high during spin testing that ordinary materials are inappropriate. All materials for spin tooling should be of “rotational” quality, free of all flaws and should have been x-rayed and certified prior to machining. Balancing: Before spin testing each assembly must be balanced to ensure that the test speed can be achieved without undue vibration. In order to balance the assembly, the tooling must be designed such that there are two journal surfaces on which to mount into a balance machine. These two surfaces must be highly precise in order to prevent wobble during the balance process. The concentricities of these journals is usually within 0.0002 inches TIR in order to achieve a low unbalance level. Hollow Arbors: Often customers will want to instrument a rotor during the actual test. This is common in Low Cycle Fatigue tests. We can design and machine holes up through both the arbor and spindle in order to accomodate the sensor wires leading from the DUT to a motion translation box called a Slip Ring. OTHER CONSIDERATIONSSpecial Capabilities: Test Devices specializes in High Speed Rotational testing and has developed a very robust in-house capability for all the tooling used in rotational tests. Very special arbors such as our designs that use hydraulics, are used to remove the arbor from the DUT without scratching polished bearing surfaces for such things as magnetic bearings, etc. Risks: Test Devices puts great emphasis on tooling that it produces for both its own use and for tooling supplied to customers for use in their own spin systems. Poorly conceived and / or manufactured tooling risks inaccurate test results; aborted tests due to arbor problems; or arbor failures that cause the DUT to fail. Failure of a spin arbor causes the loss of the test piece, results not only in the cost of replacing damaged spin tooling, test equipment, and the need to repeat the test, but worse, the loss of an expensive and perhaps irreplaceable prototype test article. Customer involvement: It is very helpful when customers can supply information prior to the tooling design. For example, decisions need to be made by the customer on how to balance the part, whether a flange should be installed on the arbor so that weights can be added or whether material can be removed from the DUT (their part). A specification on how much and where material can be removed is also helpfu. Drawings with machining data for pick up reference as well as solid CAD models with mass properties streamline the tooling design process reducing the customer’s costs. SUMMARYTooling is one of the most important aspects of a good spin test. Details addressed beforehand lower overall costs, maintain schedule, reduce risk and provide the most accurate results. Test Devices has been designing and manufacturing spin tooling with many years of experience and has developed unique and innovative ideas and processes to solve the most difficult problems. We have check lists developed from a wide range of projects that assures that nothing is overlooked and the most cost effective tooling is produced. Our arbor designs range from the routine to the extremely complex and are available to our customers regardless of test location or equipment type. Customers can use the tooling at our facility, at their facility on their own equipment or at a third party facility. Test Devices has its own in-house capability to provide not only high quality but quick turnaround. We welcome any opportunity to quote your specific needs. DEFINITIONSArbor – a fixture that holds the DUT and transfers the rotation to it. Arbor have many shapes and sizes and are usually customized to a particular DUT. It is essentially a device whose axis is inline with the axis of the DUT. Spindle – a standardized drive shaft (quill shaft) that connects the Spin System Drive mechanism (normally an air turbine) to the Arbor. DUT – Device Under Test. Normally, the customer’s rotor. Rotor – Any rotating part. Usually refers to the Customer’s part submitted for testing. Used interchangeably with DUT. Mandrel – a specific type of arbor where a shaft fits inside the hub of the DUT. The word mandrel is not commonly used and generally referred to more generically as an arbor. Pre-Load – Used generally with screw threads, it represents the amount of clamping pressure exerted by the face of fastener to the object being clamped. While there is a cause and effect relationship between the torque of a fastener and its clamping pressure, it is not a direct relationship and is dependent on the materials used, lubrication, galling, etc. Critical Speed – The resonant bending modes of a rotating assembly resulting from its natural distribution of mass and stiffness (mass / stiffness matrix). For more information, please contact one of our sales engineers at: |
|
|
Copyright© 2003, 2004, 2005, 2006, 2007 Test Devices Inc | 571 Main Street | Hudson, MA 01749-3035 U.S.A. |