As mentioned before, axial thread interference is occasionally used, mainly as a way to improve the distribution of load between threads, but it has the happy coincidence of introducing some extra friction into the relationship, which is useful when trying to avoid bolts coming undone. This practice is nearly always used for very highly loaded fasteners.

Also mentioned was that radial interference on threads is not common; that is to say, ‘bulk’ interference over the entire length of the thread is not often used. It gives an inconsistent level of additional friction during tightening, which is very unhelpful when trying to achieve a known level of load in a fastener. However, interference fits on a single or small number of threads is used, as this approach gives a constant and repeatable amount of extra friction when tightening. This extra friction can be measured using a torque wrench or electronic device, and can be accounted for in the torque-tension relationship.

Thin-walled metallic nuts are commonly supplied with specially deformed threads that act to give a controlled and constant amount of friction. Originally designed for use in aerospace applications, these are widely used in motorsport. Most people tend to call them K-nuts, and they come in various types. The locking action is achieved by the elliptical deformation, and the most commonly used are the hexagonal K-nuts with a flange, which also exist in a high-temperature silver-plated version. Other types that are riveted to sheet metalwork or composite panels provide thread-locking in applications such as engine intakes.

All-metal locking nuts are a real favourite in aerospace applications, so we also find them used in motorsport. Another popular type of nut, often known as aero nuts or aerotight nuts, have slots cut towards the top of the nut, a couple of threads down from the top, de-stiffening the top part of the nut which is then deformed slightly in the axial direction.

Various suppliers of wire thread inserts market inserts with a single deformed thread to provide a constant amount of friction when installing a fastener. Other than chemical thread-locking methods, deformed wire inserts are the main method of positively locking a standard fastener into a threaded hole, as they have the advantage of providing a constant amount of additional friction to the system. Chemical thread-locking methods are prone to inconsistencies between operators and differences in application, speed of tightening and so on, and here the mechanical wire insert is much more predictable, albeit at a much higher cost.

Locking thread inserts are sometimes installed in nuts that are made of soft materials or those prone to surface damage during tightening: titanium and aluminium nuts are available in this locking configuration. The wire insert in the nut also provides an improved load distribution between threads, leading to a lower stress concentration factor on the male fastener.

Written by Wayne Ward

It is possible to specify fasteners with a dry threadlock compound applied. In such compounds, an adhesive is encapsulated in a resin binder, typically epoxy. The encapsulated adhesive is not released until sufficient pressure is applied, and this typically happens as the pre-load is developed. The prevailing torque during tightening – that is, the extra torque due to friction – for this type of threadlocker is due simply to the increased drag from the epoxy. Unlike wet threadlockers, there is no torque developed from an unknown and increasing amount of threadlocking compound beginning to cure and set.

These micro-encapsulated adhesive threadlocking systems are available in a variety of specifications, depending on the application and the service temperature, and are applied over a specified number of threads. The recommendation is that they are not re-used, as the holding power on the second and subsequent uses is a function only of the remaining adhesives after the first use. Some compounds are designed to provide a seal in addition to locking the thread.

Mechanical thread locking can take a number of forms, but a convenient starting point is the type which is most similar to the dry threadlocking compounds. It is also possible to specify fasteners that have a number of threads coated with a thin layer or patch of polymer, which acts to put the fastener into slight interference in its hole. There are also options as to the amount of coverage, which will dictate the prevailing torque measured on installation. Nylon is typically used as the locking material.

Continuing with polymeric locking elements, some specialised fasteners are supplied with a polymer bung inserted radially into the thread. As with the previous option of a nylon patch, any prevailing torque is constant after initial compression of the polymer element.

Interference-fit threads, where the male and female components are in radial interference, are not commonly used owing to the unpredictability of the prevailing torque and the possibility of thread damage. Small differences in the amount of interference can make large differences in prevailing torque; however, a degree of axial interference achieved through the use of deliberately mismatched thread pitch is sometimes used on very critical high-strength fasteners. This provides a degree of protection against loosening, and has been shown to improve the distribution of load between the individual threads and to markedly reduce the degree of stress concentration at the first engaged thread.

Written by Wayne Ward

The best way to ensure that a fastener remains tightly fastened is to understand its behaviour, and to tighten it accurately. The most critical fastener in any reciprocating engine with split con rods are the con rod bolts, and these are rarely equipped with any positive method to prevent loosening. However, con rods are a rare case where we can accurately measure pre-load (indirectly via measurements of bolt extension and a knowledge of the stiffness of the fastener).

For most other applications we do not have this luxury. There are many ways of preventing a fastener from loosening; some are chemical and some are mechanical. The chemical solutions to fastener tightening are based on either wet or dry thread-locking compounds. There are various dry thread-locking compounds, but they aren’t very widely used in comparison with the wet thread-locking chemicals that many of us are probably familiar with. They come in various strengths and levels of temperature resistance. The weakest is specified for fasteners that require routine disassembly, while the strongest types are really only used for applications where the intention is for the fastener never to be removed. The very strong grades also often require heat to weaken them before studs can be removed.

Such thread-locking compounds are basically anaerobic adhesives, which rely on an absence of air in order for them to cure. Their cure time is affected by temperature, and although there are chemical accelerants available to reduce cure times, the final strength of the thread-locking compound is often compromised by their use, although new activators have been developed by some companies that bring the cured strength back to 100% of the ‘unactivated’ strength.

The strength of the bond is also affected by the material to which the compound is applied. When applied to a bare steel substrate, such compounds are often much stronger than when applied to fasteners that have been chromated, for example. The bond gap also has a large effect on the cure time and the final cured strength of the thread-locking compound, with larger gaps showing increased cure times and significantly diminished performance, as measured by the torque required to loosen the fastener.

Experience of using thread-locking compounds in trials where clamp load was measured for a critical application shows it is important to develop a process for the use of the compound, detailing how much is used, where it is applied, how much delay there is before fitting the bolt and how it is tightened. During some trials I organised, the clamp load – particularly with high-strength thread-locking compounds – showed much greater variation for a given torque than for oil-lubricated fasteners. It was clear from the trials that the thread-locker was beginning to cure and to exert some resistance before the joint was fully tight.

Written by Wayne Ward

One fundamental choice that we need to make is that of threadform. We need to carefully select the correct size of fastener so that it can provide sufficient pre-load and withstand the service loads we expect to subject it to. There is a wide variety of threadforms to choose from; not all threads are designed for use as fasteners, so those such as acme and trapezoidal threads can be discounted instantly. We will generally choose between metric (M) and unified (UN) threadforms, both of which have flank angles of 60°. Unified threads are imperial (inch) sizes and are most widely used in the US, which continues to use the inch as its preferred unit of length.

There are fatigue-resistant versions of both these types of threads, known as J-form threads. For any given thread pitch, these have a more generous root radius on the male thread than the ‘non-J’ equivalent. The increase in root radius increases the fatigue strength of the fastener for two reasons. The first is that the stress concentration is reduced slightly owing to the radius increase, and the second is that the minor thread diameter is also increased. J-form threads are denoted by MJ for metric and UNJ for imperial sizes, and there are only certain combinations of nominal diameter and pitch for which J-form threads are available.

Whitworth threads incorporate a controlled radius on the major diameter of the tap, so that the female threaded component is rendered more resistant to fatigue by having a reduced stress concentration at its major diameter.

The aero thread is a very clever but complex thread system that has never gained widespread acceptance. It incorporates a number of features that make it resistant to fatigue but it is rarely (if ever) used for new designs, and I have never seen an example of this in anything other than a textbook. The male thread is semicircular and relatively shallow. It therefore resists fatigue for the same reasons as the J-form threads but to an even greater extent.

The female thread is cut into a nut or casting, for example, and is of a completely different form to the male thread, being similar in terms of geometry to a common 60° flank angle thread. There is an intermediate member which is a thread insert to be installed into the female component. The thread insert has a 60° flank angle thread on the outside and the semicircular thread on the inside. Thread inserts de-stiffen the female thread, and this is known to improve the distribution of load along the thread. In a conventional metric or unified thread, the first loaded thread takes very much more of the load than any of the other threads, and the use of female components of lower stiffness than the male thread makes the load distribution much more even.

Aero threads are expensive to produce, and the lack of availability of suitable inserts makes them impractical. However, the concept of using thread inserts to improve load distribution (and hence reduce stress concentration factor or improve fatigue strength) is very valid, and suitable thread inserts are available in most metric and imperial thread sizes.

Written by Wayne Ward

It does not take a huge leap of imagination to look at the actual applications of helical springs as fasteners. There are applications where the imposed displacements of various components are large, and the amount of compliance required in the joints is very large. In such situations, using a threaded fastener becomes difficult owing to its stiffness. We need something that still maintains a controlled load, but over a much larger range of joint displacements.

The main application of helical springs as fasteners in motorsport is in exhausts. It can be difficult to manufacture the various separate pipes with flanges or lugs that are accurate enough to use traditional bolted joints. Radial lugs on adjacent pipes are sometimes joined using short ‘kinked’ plates that have their own compliance, along with threaded fasteners.

However, it can be simpler and more convenient in practice to use extension springs to pull adjacent parts of an exhaust system together. The requirements in terms of the accurate positioning of the lugs or loops between which the springs are stretched are less exacting than for joints using plates and threaded fasteners. If you are in the position where you need to quickly disassemble and then reassemble part of a hot exhaust, you may find it difficult to wield spanners and ratchets. A simple spring puller may be easier, faster and more pleasant to use in such situations.

The problem with using springs to pull a joint together on an engine where a range of frequencies are generated during service is that a spring will tend to resonate when excited by a number of frequencies of vibration. Once this happens then the stress in the spring is likely to lead to premature failure. It is a similar situation to spring surge in valve springs, although in this case the spring is an extension spring rather than a compression spring.

I have seen a couple of methods to counteract the effects of resonance, on motorcycles used for 24-hour endurance racing. The first is simply to run a ‘bead’ of high-temperature silicone along the spring. It works but looks messy, and the silicone bead can fall off if adhesion is poor. The second method is to use polymer heat-shrink over the entire length of the extending portion of the spring. This is more reliable, as the heat-shrink is essentially in tension around the spring and is very unlikely to become detached; it also looks much better than using silicone sealer. However, service temperatures need to be borne in mind because if the heat-shrink is too close to a hot exhaust then it will melt or burn and will no longer offer any damping.

Written by Wayne Ward

However, there are conditions under which the fastener itself can cause significant damage to its adjacent components. In the presence of an electrically conductive liquid, for example, dissimilar metals in contact with each other can behave in such a way as to accelerate the corrosion of the most chemically reactive of them.

There is a ‘pecking order’ for the reactivity of metals, a property known as nobility. Metals that are very slow to react, such as gold, are termed noble metals, while those that are more prone to react are known as ignoble or corroding metals. Two metals with similar nobility, in contact with each other and an electrically conductive fluid, may not pose a problem – in fact, a problem may not occur with a pair of metals such as steel and aluminium. However, if the steel is substituted for stainless steel, or the aluminium is changed for a magnesium part, there may be serious corrosion of the less noble component.

The mechanism by which this happens is known as galvanic corrosion, and the nobility of each metal can be quantified by something known as electrode potential. Gold has the highest electrode potential, of +1.5 V, while the lowest electrode potential of any metal that we might use in the construction of an engine is magnesium, at -2.37 V. The material with the lower electrode potential of any pair will corrode, and the speed at which that happens is affected by the magnitude of the potential difference.

The potential difference in the case of gold and magnesium is 3.87 V, so it would be a very bad idea to gold-plate a steel fastener that would be in contact with magnesium in submerged or humid conditions. The electrode potential of the unplated steel fastener would be around -0.4 V, giving a potential difference with magnesium of -1.97 V; the gold plating increases the potential difference by a factor of almost two. Stainless steels are far more electropositive than non-stainless steels, and so can cause much faster corrosion of aluminium and magnesium than non-stainless steels.

Thankfully there are only a few instances where this is likely to affect a race engine, and these are generally restricted to fasteners in contact with the water cooling circuit. Some engines have fasteners passing through the coolant passages of the cylinder block, and the same can happen in the water pumps or elsewhere in the coolant circuit.

There is a balance to be struck – the engineer needs to protect against the corrosion of the fastener, but must also be wary of inadvertently accelerating the corrosion of the cylinder block or pump housing. Using deionised water rather than tap water can be an effective way to reduce the effect of galvanic corrosion while the water remains uncontaminated and hasn’t dissolved too many ‘nasties’. Also, corrosion inhibitors need to be used with caution, as too much inhibitor can make the coolant more electrically conductive and, perversely, accelerate corrosion.

Written by Wayne Ward

The amount of radial thread engagement is referenced to the pitch of the thread and the nominal diameter, and is controlled by varying the inner diameter of the female thread. It is usual to provide a female thread that is weaker in terms of shear strength than the male – in this case the nut will strip, rather than the thread on the bolt. Provided that the major (outside) diameter of the male thread remains at a certain size, the load required to strip the female thread should be unaffected.

However, this assumes that the male and female thread axes remain parallel and concentric, and this is not always the case. Where there is excessive lateral play (clearance) in a thread, the result can be that the axes are not coincident. In this case, the assumptions about thread-stripping strength are no longer valid, and if too large a pilot drill is used along with a male thread towards to lower end of the tolerance range for thread class, the thread axes can become so far displaced from one another that they become disengaged on one side.

The formula used to define radial thread engagement calculates maximum possible engagement height based on the pitch of the thread and the thread angle. If the male and female threads were perfectly sharp and were an exact fit, 100% engagement height would be the pitch x cos 30° (the semi-angle of the thread). To allow for a radius root in the male thread, 60° threads allow the female thread to be truncated by 25% of the engagement height – that is, the maximum theoretical engagement is 0.75 x 0.866 x pitch (cos 30° = 0.866) = 0.6495P

The percentage engagement is the ratio of the actual overlap of the thread radially to this 0.6495P, which is:

% radial engagement = 100 x (nominal radius – pilot hole radius) ÷ 0.6495P

It is easier to calculate by multiplying the top and bottom of the formula by 2, so that we then have the calculation in terms of diameter, for which we will have data to hand, and this is the formula you might find in texts on the subject:

% radial engagement = 100 x (nominal diameter – pilot hole diameter) ÷ 1.299P

For an example of M10 x 1.25P, the nominal diameter is 10 mm, the pilot is 8.8 mm and the pitch is 1.25mm Therefore the percentage engagement is 100 x (10 – 8.8) ÷ (1.299 x 1.25) = 73.9%

The figure of around 75% engagement is typical for 60° threads, and small changes make a large difference to the result. In the previous example, if the pilot hole is adjusted to 8.9 mm, the engagement changes to 67.7%

In production engineering there is an incentive to use a lower thread engagement, and figures as low as 50% are typical. Two reasons for this are lower production costs and faster production, as the amount of metal removal is reduced, as is the tapping torque.

There are reasons for us to be wary of this practice though. First, it increases the effective contact diameter of the threads, and will give a higher tightening torque for a given load. If we have specified a tightening procedure based on the assumption of 75% engagement, we may have less pre-load than expected. Second, by increasing the pilot hole diameter, we are decreasing the shear area of the male thread, making it more prone to strip.

Written by Wayne Ward

Compared to an engine block, a chassis is a relatively straightforward component – it is physically large, and the load paths are well defined and less complex than in an engine. There are some real difficulties in the manufacture of complex structural components, though, and it definitely isn't an easy task to make a successful composite engine block, for example. Part of the development bottleneck is due to restrictive regulations in some race series.

However, even if regulations were free, would we see composites supplanting conventional materials for blocks and heads? Only time will tell.

There are difficulties where we want to join one component to another, especially if we want to disassemble these parts at a later date. Bolted joints can be a real pain, and unless fastener loads are very low, it is not a good idea to try to put a thread directly into a fibre-reinforced composite component. Cut threads work badly, and moulded threads are often simply formed from weak resin.

Where we want a female thread, we need to provide a metallic ‘hard point’ in the composite, and this means bonding in or, more likely, moulding a metallic component in place. It should theoretically be possible to mould a hard point into a composite component with the thread already cut, but often this is not successful.

Experience often leads people to cut the thread after the hard point has been moulded in place. It isn’t usually prudent to rely solely on the resin to react the fastener torque and service loads, as the hard point needs to consider loading conditions and should provide some mechanical design features to prevent unwanted movement. The hard point itself can become quite bulky, especially as we need to have a certain amount of composite around these components – it can place the fastener axis far enough away from the applied load such that fastener bending stresses need to be more carefully considered.

Where a male thread is required there is a huge variety of off-the-shelf fastener components that combine a male thread with some unusual head geometry. The head of the fastener is not used to pre-load the fastener – the head is ‘buried’ within the composite and is a locking feature – and the heads of such fasteners are rarely axisymmetric. Polygon-shaped heads or discs with holes provide the opportunity for the composite to mould the head firmly in position.

The same type of fastener is also sometimes used to provide a female thread – here, a larger-diameter spigot is tapped rather than being externally threaded. However, because the fastener’s head is physically large, such fasteners don’t lend themselves to many applications where we might want a thread in a highly stressed engine component.

Written by Wayne Ward

All of this electric power requires high voltages – if we wanted to use 60,000 W of electric power at a conventional 12 V, for example, we would need 5000 A of current. Instead though, currents are kept to much more sensible levels, and for voltages the published literature from one component supplier for Formula One KERS states 500V, although the actual figure may be higher.

The advantage of a high voltage is that lower currents are required, and this means that the cross-sectional areas of conductors are reduced. The disadvantage is that high voltages like to jump across air gaps or creep along surfaces to conductors at lower potential. This means we need to leave a certain minimum air gap between any component at high voltage and one at lower voltage or connected to earth. Where two such components both touch the surface of an insulator connecting them, the problem is worse. If the air gap is respected, there is a larger distance across which charge will creep if there is a solid insulator.

Together, the effects of charge jumping a gap or travelling along the face of a conductor are known as creepage and clearance. Clearance is a fixed distance per voltage difference, given a known medium (for example air) over which the potential difference exists. For air this is 3000 V/mm, although it depends on temperature, air density and humidity. In any case, it is relatively easy to calculate and to leave a sensible safety factor.

Creepage is a different matter though, as it depends on the insulating material connecting the two conductors, how contaminated the surface is and what the contaminants are. The main area where creepage and clearance are likely to be a concern is the power electronics unit, because of the large number of high-voltage components in close proximity, but the same concerns exist in the motor, battery and voltage converter, if one is used.

Fasteners can often be the problem. We need them to hold lots of things together, but they can form an inconvenient conductive link between high- and low-voltage components. Where a non-conducting fastener can be found of acceptable strength, this is often a good solution. We should not confuse such fasteners though with nylon ‘registration plate bolts’ that we might find on passenger cars; there are non-conducting fasteners made from a number of high-strength polymers such as PEI and PEEK. These have a useful degree of strength but poor creep characteristics, so we need to make sure that the combination of service temperature and strength doesn’t cause them to lose load in service.

Of course, we have the option of using metallic fasteners and maintaining acceptable gaps where required, but this necessarily results in larger circuit boards that then need larger enclosures. Suddenly we have a physically larger and heavier hybrid component that tends to be more difficult to package and invites a lot of disparaging comments from the chassis men. In some cases it is possible to shroud conductive fasteners so that creepage and clearance effects are avoided, but often the best solution is a non-conducting bolt or stud.

Written by Wayne Ward

The fastening of thin components is not technically difficult to achieve: we can use the same type of fasteners that we use in normal circumstances, but we need to be aware of the difference in behaviour of threaded fasteners, which can tend to lead to fastener loosening or accelerated failure. In the case of airboxes being fastened to cylinder heads with threaded fasteners, loose fasteners or bolt heads that have fallen off due to fatigue can be swallowed by the intakes and cause a catastrophic engine failure.

As readers may already appreciate from reading previous RET-Monitor articles on fasteners or technical papers on the subject, most of the load is taken by the first few threads closest to the applied load. In the case of fastening thin components, this means that the clamped length of the joint is much shorter than normal. This has two effects, neither of which help towards the success of the joint.

The first effect is to increase the internal load coefficient, which dictates the fraction of the applied service load to which the fastener is subjected. The internal load coefficient increases as the ratio of fastener stiffness to joint stiffness; in simple geometry the joint stiffness is calculated using the ideal stiffness of truncated cones. (For a recap on this you may find it useful to refer to previous articles or mechanical engineering/fastener textbooks.) For thin joints, the stiffness of the joint is lower than usual compared to the stiffness of the bolt. This leads to an increased load on the bolt due to cyclic service loads, and lower fatigue life. Using a washer can improve this situation a little.

The second effect is that the fastener itself physically stretches very little in developing the pre-load, so any slight relaxation of fastener ‘stretch’ (elongation) due to the joint ‘settling’ can easily lose a significant proportion of the desired pre-load. If the cyclic service loads are sufficient to then allow joint separation, the fastener itself is subjected to 100% of the applied load.

Unless the fastener is sized for this – which is certainly not the case in any optimised joint – you are likely to find either broken fasteners (with fracture surfaces characterised by fatigue marks), loose fasteners or missing fasteners. Again though, use of a washer is a good way to improve the situation. This increases the working length of the bolt, which increases the stretch in the bolt (for a given strain, the stretch is in direct proportion to the working length of the fastener).

Where two thin components are being fastened together, you should consider using washers on both sides of the joint. Where a thin component is being fastened to something more substantial that’s provided with a tapped hole, counter-boring or counter-drilling the tapped hole again increases the working length of the fastener.

Written by Wayne Ward

In terms of their tightening, slender bolts and studs behave in the same as any other bolt. For example, with identical dimensions, surface finish and coefficients of friction, a stout bolt and a slender bolt should develop the same pre-load for a given torque.

In service, however, the slender bolt is much more likely to be affected by resonance. This is a condition where the bolt is excited and vibrates at one of its natural frequencies by a stimulus coming from the engine. There are all kinds of vibrations in an engine, and these have various harmonics acting at multiples of the fundamental speed of the component or system generating them. Slender fasteners are more likely to be affected because their natural frequencies are lower than those of stouter fasteners. When a resonant condition is reached, the maximum amplitudes in a slender fastener are greater than for a less slender one.

The life of a fastener may be reduced significantly owing to the higher stresses encountered. The resonant condition will usually be one of bending, and the fastener will often show signs of bending fatigue at the points at which it is restrained, which is often in those parts of the threaded portions of the bolt which are most highly stressed in service.

Fortunately there are a number of easy remedies to this problem. These revolve around increasing the natural frequency of the fastener, which can easily be done, or by limiting the amplitude of vibration and hence lowering the stresses involved.

There are a number of ways to change the natural frequency of the shank. It can be increased markedly by even a slight increase in the shank diameter. Changing the length of the shank is also effective; for waisted (reduced diameter) shanks this can be done by increasing the length of the threaded portions on the fastener (decreasing the length of the waisted section) or adding an increased diameter section close to the ends of the fastener.

Another way to change the natural frequency is to add a location diameter part of the way down the shank. This effectively creates two short shanks on the same fastener, increasing the natural frequency significantly. In practice, however, it is difficult to guarantee location. In order to get the parts to go together, it is far more likely that a clearance condition between the location and the hole will be created. In this case, what will happen is that the natural frequency can actually be reduced slightly, owing to the effective addition of a mass to the slender beam. What will happen is that the amplitude of vibration and hence the stress will be reduced.

Written by Wayne Ward

The increasing use of boosted engines in motorsport, especially turbocharged engines, means that exhaust temperatures are much hotter than usual, often necessitating more exotic exhaust materials. Where there are applications for fasteners in such exhaust systems, we therefore need to consider the use of something more exotic than a conventional tempered steel fastener; while these are generally excellent quality, they are designed only for use at moderate temperatures.

Austenitic stainless steel is less affected by heat but starts off at a lower strength level, and a lot of commercially available fasteners have poor-quality threads. We probably won’t want to consider off-the-shelf austenitic steel fasteners for highly stressed, high-temperature applications.

Martensitic stainless and duplex stainless grades are capable of being hardened to a high-strength condition, and so are used where conventional tempered steels are unsuitable and where austenitic grades are not strong enough. Fasteners in martensitic and duplex stainless materials are custom-made only.

Titanium alloys can be used for higher temperature applications than conventional steels, but the cost of titanium fasteners often makes them difficult to justify when bespoke designs are required. However, they are widely available commercially in a variety of bolt styles, from cap-head screws to bolts with flanged heads. So, if you can find the correct fastener for your application, the cost may not be as bad as you might fear.

If we are considering the use of threaded fasteners for high temperatures (possibly 600 C or higher), we should start to consider superalloys. There are three common types of superalloy, based on iron, nickel and chromium. Such materials are generally quite expensive though, owing to the high proportion of costly alloying elements and the expense of processing.

Materials such as Inconel and Nimonic alloys are the same types of materials that we might consider for other ‘hot’ components such as turbine wheels or exhaust valves; they suffer very little loss of properties at high temperatures relative to ‘normal’ fastener steels. Without special processing though, they may not offer very high strength at room temperature, but it is the level of strength available at working temperatures that will interest us for fasteners used on the exhaust system of a boosted engine. We might normally see superalloys such as Nimonics (sometimes used for exhaust valves) being used for fasteners and Inconel alloys as sheet metal being turned into exhaust systems for teams with large budgets.

Written by Wayne Ward

The design engineer might be mistaken for thinking that, having designed a stud from the strongest available material, having incorporated the most fatigue-resistant thread forms, taken great care over other design features and used the best manufacturing methods, he or she has done everything in their power to make the stud as reliable as possible. It is true to say that they have gone a long way towards this goal, but the nut also has a very important part to play in the endurance of the stud.

Even for a completely fixed nut geometry, there is one important factor over which the design engineer has control and which can lead to a significant improvement in stud fatigue resistance. This ‘magic’ factor is the material from which the nut is made. The design of the nut can also have a strong effect on the stress concentration factor experienced by the bolt or stud, but for this article we will look at the effect of material.

In order to understand this, let us reflect on the lesson first published in 1902 by Russian scientist Zhukovski, who postulated that the loads borne by threads in a nut were not evenly distributed, as we might imagine, but the threads closest to the load application bear a disproportionate share of the load. The distribution was shown to be a hyperbolic relationship depending on several factors. However, for the example of a steel nut and bolt with ten engaged threads, Zhukovski showed that the vast majority of the load was taken by the first few threads, with 34% taken by the first complete thread.

Since Zhukovski’s time, calculation methods and the computing power needed to carry them out have advanced, and more recent studies have put the stress concentration due to the first thread higher than Zhukovski’s figure. Several of these studies have taken a standard example of an M10 fastener with five complete engaged threads; both simulation and measurement show that the first thread takes between 36% and 41% of the total load.

The key to improving the load distribution in the threads is to make the nut threads less stiff. By doing this, the load is more easily transferred between threads by both shear and tension effects. The practical way to do this, for a nut of fixed geometry, is to change the material of the nut to one with lower elastic modulus. For nuts with a lower elastic modulus than that of the stud or bolt, the stress distribution is made more even than Zhukovski’s prediction, giving a more favourable stress concentration factor, but where the nut material’s modulus is greater then the load distribution will be worse, leading to a more severe stress concentration.

Written by Wayne Ward

As is widely explained in general mechanical engineering textbooks or specialist fastener books, the amount of the cyclic service load borne by the fastener can be predicted if we know the stiffness of the fastener and the stiffness of the clamped members in a joint. The basics were explained in some RET-Monitor articles first posted in 2009. The ‘Back to Basics Part 2’ article gave the formula for the load coefficient, the calculation of which tells us the proportion of the cyclic service load the fastener will be subjected to.

If you don’t want to subject yourself to some basic maths, we can state one inalienable fact – for a given fastener stiffness, the cyclic service load will be reduced if the stiffness of the clamped members is increased, providing that the joint has sufficient pre-load to resist joint separation. For fixed clamped member geometry, it is unlikely that a change of material to anything stiffer – as in having a greater elastic modulus – will be possible. Not only is it usually impractical, it will often involve far heavier clamped members.

A stiff washer of increased diameter compared to the original part which tightened onto the clamped members (whether this was a nut or washer) will clamp a greater volume of material, and the stiffness of the clamped members will increase as a result. Only in very simple geometry can the effect be easily predicted using ‘pencil and paper’ maths. Engines in general are full of complex structural parts, so the clamped members cannot easily be approximated mathematically by a series of tubes or hollow frusta (truncated cones).

Where geometry is simple enough to be modelled using simple geometric forms, the stiffness of a clamped joint member is usually thought to be most accurately modelled by a frustum rather than a tube. The stiffness of a frustum is a little more complex to calculate than that of a tube; load is known to ‘spread’ to clamp a greater volume than a simple cylinder where material exists for it to spread into. A cone ‘semi-angle’ of 30° is often used, having been adopted from some early studies into joint stiffness (see references), although various studies propose other values.

By clamping a greater volume of material through increasing the outer diameter of the clamped members, the washer leads to a decrease of cyclic load on the bolt or stud. The washer needs to be sufficiently stiff to be able to transmit load to a greater diameter than the head of the bolt or nut which loads it. Thin, flexible washers are of little use for this purpose.

Washers also promote reliability by helping to protect the face of castings or other components from directly having to withstand highly loaded sliding contact as the fastener is tightened.

References

1. Little, R.E., “Bolted Joints: How Much Give?”, Machine Design, November 1967

2. Osgood, C.C., “Saving Weight on Bolted Joints”, Machine Design, October 1979

Written by Wayne Ward

In this article though we are not going to deal with round wire circlips, as commonly used for retaining piston pins; they were discussed in a recent piston article. Here we will deal with the more common flat circlip, and look at some specific variations that can be useful.

Circlips are inexpensive components designed to be elastically deformed to allow them to fit into specially machined grooves provided in shafts or housings. The amount of deformation required for them to fit requires that they are made from materials with a certain amount of elastic deformation. Moreover, they operate in an elastically strained condition, and the amount of radial load must not diminish markedly in service.

Circlips are generally mass-produced by stamping from sheet metal, and are usually steel, although some are available in other materials such as stainless steel or beryllium copper. Other manufacturing techniques such as machining or laser cutting can be used, and lend themselves to special designs where required.

Most circlips are of a standard pattern, although for racing there are some interesting variations on this. For rotating assemblies, where a circlip is used on the rotating parts, a ‘balanced lug’ circlip may be used, which is designed to be ‘in balance’ when fitted – that is, the centre of gravity of the circlip is coincident with the axis of the hole into which it is fitted. That is not true for standard circlips and, depending on the speed of rotation, significant out-of-balance forces may occur, requiring the assembly to be balanced to take account of the circlip out-of-balance forces. This isn’t a problem in itself but, if the standard circlip is ever removed and refitted, it does mean the circlip has to be fitted in precisely the same orientation as before, to avoid putting the assembly out of balance.

Bowed circlips are a deliberate attempt to provide a degree of pre-load and to take up small amounts of axial play in an assembly. In its free state the circlip has a precise amount of curvature so that, when flattened or partially flattened in its installed state, it provides a load on the adjacent components.

Bevelled circlips are a variation on a standard flat circlip. A shallow angled bevel is applied to the outer diameter on one side of the circlip (assuming that this is an internally fitting circlip) by careful sizing and positioning of the groove; the circlip does not go fully ‘home’ in the groove but is forced axially, owing to the reaction between the circlip bevel and the corresponding surface of the circlip groove, which has an angle equal to that of the circlip bevel. The bevelled circlip has the advantage over a bowed circlip of a much higher stiffness, as it doesn’t rely on elastic axial deflection of the circlip to take up any axial play.

Fig. 1 - Circlips come in many shapes and sizes, with some designed for specific tasks

Written by Wayne Ward

Greater attention is paid to the production methods for male fasteners, as they are generally more apt to fail than their female counterparts. Machine-cut threads on highly stressed cyclically loaded male fasteners are generally frowned upon, as their durability compared to rolled threads is notoriously inferior. For many years, when it came to producing female threads in metals, there really was only one game in town, and that was to use a conventional thread tap. Before tapping, a hole is drilled at the size of the minor diameter of the internal thread, and the ’space’ into which the fastener is screwed is created by the tap, which essentially has the same form as bolt, but with multiple cutting edges.

There are two more modern methods which each have their own advantages. The first is internal thread milling. A milling cutter with the correct thread form and a given pitch, used in a CNC milling machine, is capable of producing any diameter of thread and either hand of thread of that specific pitch. The range of threading tools required to cover many different sizes of thread is therefore smaller than the number of taps required for the same job. It is also possible to get the full thread form closer to the bottom of a blind hole using thread milling than is normal with a conventional tap.

The second method is internal thread forming, which is closer to the conventional tapping process than thread milling. It requires a tool for each combination of diameter and thread pitch, as is the case with conventional thread tapping. However, the tool has no cutting edges and there is no cutting of metal involved in the process. The pilot hole for the thread forming process is larger than for conventional tapping, as the material has to ‘flow’ inwards to produce the full thread form. The process forms a characteristic ‘claw’ shape at the internal diameter, but it is at the major diameter of the internal thread that the advantages arise. The grain flow here is favourable, strengthening the material and making the internal thread more resistant to fatigue loads.

Forming internal threads (often called roll tapping) has some advantages in terms of tool longevity, but there are also some disadvantages. The process is less suitable than conventional tapping for materials with limited ductility, although this has improved with tooling design over recent years. Polymer materials which exhibit high elastic strains also tend not to perform well with thread-forming taps. The method also tends to displace material axially at the entrance to a hole. Countersinking or counterdrilling of roll-tapped holes is generally necessary to remove the displaced metal at the hole entrance, although thread-forming taps are available with features which automatically remove the offending material.

Written by Wayne Ward

There is a phenomenon called nut dilation at play to some extent in all female threaded components. As the name suggests, nut dilation is concerned with an increase in size of the nut brought about through the act of tightening the fastener, although the effect is not confined to nuts alone but to all female threaded parts.

A normal thread has a definite flank angle and, as we tighten the fastener, the natural behaviour of the parts in contact acts to move the nut material along the 'ramp' of the thread flank. The female part is generally the less stiff part in the radial sense, hence the effect of dilation.

The strength of the female thread is affected by nut dilation in terms of its resistance to thread stripping under load. As thread dilation moves the female thread radially outwards (and to a much lower extent the male thread is compressed radially inwards), the 'overlap' of the inner and outer threads is reduced. As the female thread dilates, the shear area of the female thread at the maximum contact diameter is reduced. Where the stress exceeds the shear strength of the nut material in shear, the female thread will strip, assuming that the male thread is strong enough to make the female thread strip first. The same applies to the male thread in the case that it is likely to strip first, except that the shear area is calculated at the inner diameter of the female thread. Whether it is the male or female thread which strips first, nut dilation means this undesirable result will happen at a lower than expected load.

Nut dilation also affects the torque-tension relationship in two ways. As the nut dilates, the effective thread contact diameter also increases, leading to an increase in the thread friction term of the equation. Additionally, the nut underhead face is also increased in terms of mean diameter, and this means that underhead friction is correspondingly increased too.

There are a number of ways to counter the effects of nut dilation, but all generally mean an increase in component mass. The aim is to increase the radial stiffness of the component. One might decide to increase the height of the nut or the length of the engaged thread. However, as the load is taken preferentially by the first few threads of the joint, this has little effect. Increasing the across-flats dimension of the nut - or the amount of material surrounding a female thread in a machined part - is effective, as is choosing a material of increased modulus.

Written by Wayne Ward

Con rod bolts are generally equipped with bi-hexagon (12-point) heads. Conventional sockets tighten bolts through contact close to the corners of a hexagon because of large clearances between socket and fastener. More corners equals more torque capacity without danger of the bolt becoming 'rounded'. Also, a small socket can be used to impart a given torque, requiring a smaller diameter bolt head, and a smaller bolt head means less friction, and less friction means that a lower torque is required to produce a given pre-load. Twelve-point heads can't be formed by conventional machining processes, although there are material removal methods such as spark erosion that could produce the required shape. Bolt heads with internal tightening features - socket-head bolts - are also often produced by forging.

It is commonly understood that forging offers other advantages too. Forging is a process that produces a shape by material displacement rather than by material removal. It is widely used in the production automotive engine industry for many components, most notably crankshafts, con rods and pistons. In racing, pistons are still mainly machined forgings, but forged crankshafts are more of a rarity owing to the cost of producing the tooling. Con rods occupy a middle-ground; where sufficient parts of one design are produced, forgings are a common starting point, and manufacturers will use a common forging for several similar designs. The grain elongation and flow in the forged material is reckoned to produce worthwhile improvements in fatigue strength.

For bolts, the case is less clear-cut than for some larger components, and there is certainly not universal agreement between manufacturers of forged-head bolts as to whether there is a real improvement in component strength and durability due to forging. When interviewing bolt manufacturers, I hadn't expected to find such diverse opinion. It can't be argued that the grain flow doesn't take place, or that the grain flow wouldn't be beneficial.

However, the underhead of the bolt is generally machined, and for very highly stressed bolts it is common to find that the underhead has been slightly undercut too. This serves to remove some of the forged material where it would present the greatest advantage; the machining operation also interrupts the grain flow.

Additionally, for a number of high-strength materials used only for very highly stressed bolts, the material has to be heated in order to forge the head detail. If incorrectly done, this can leave the material in the highly stressed transition between head and shank in a lower-strength condition than we would ideally like to find. As hardness and strength are generally related, finding a critical area of a bolt soft will not instil confidence.

However, for many designs of critical bolts, we can cope with a slightly softened transition between head and shank. If we take the example of a con rod bolt, it is often the case that the most highly stressed part of the fastener is the waisted shank; the underhead transition is usually larger than the shank and is therefore less highly stressed.

Written by Wayne Ward

There are advantages and disadvantages to each of the methods described. One method, which is used extensively in industry but rarely in racing, is to tighten fasteners to yield. When the fastener is in service, the load in the fastener increases only moderately with applied load until the joint begins to separate, beyond which the load and stress in the fastener begins to increase much more rapidly.

In order to avoid the onset of joint separation, high loads are called for. The highest load that can be applied by tightening is that which occurs at bolt yield. The stress in the bolt is a combination of tensile stress caused by the elongation of the bolt, and shear stress due to the torsion applied to the bolt in tightening. Therefore, yield occurs due to this combined stress rather than pure uni-axial tension. The load and extension at which this occurs is lower than the load that would cause yield in tension alone.

If we refer to Fig. 1, other tensioning methods discussed so far put us somewhere along the linear region of the blue curve, before it deviates from the red curve due to yield. As the joint is loaded or unloaded, the bolt load and extension change in line with this pre-yield load-extension curve. However, once the bolt has yielded and the torsional component is removed, when the joint is loaded the load extension curve returns to the same gradient as before. Under additional tension, the bolt will not behave as a yielded bolt until the blue curve again almost reaches the red uni-axial tension curve.

In the book by Bickford, a fatigue graph is reproduced where a bolt is tightened to a stress level of 450 MPa, 750 MPa and then to yield. At 10,000 cycles the maximum dynamic loads in the bolts at failure were 5500 N, 8000 N and 11,000 N respectively. We can see that by tightening to yield, we can increase the fatigue life of the bolt.

There are good reasons though why people don't like to use yield control. First, many people are wary of it because their entire experience is with designing parts that work in the elastic region of the stress-strain curve; plastic deformation is generally seen as a mechanical failure. Second, the equipment required to detect yield accurately and repeatably relies on algorithms such as measurement of torque gradient. Third, race engines are often rebuilt, and bolts tightened beyond yield unless specifically designed for repeated use can tend to damage threads.

Typically in industry, where 'stretch bolts' are often used, the advice is to use them only once.

Fig. 1 - The graphs shows the load extension behaviour for a fastener in uni-axial tension (red) and combined stress (blue)

Reference
Bickford, J.H., and Nassar, S., "Handbook of Bolts and Bolted Joints", Marcel Dekker, 1998, ISBN 0-8247-9977-1

Written by Wayne Ward

The method of torque tensioning remains a favourite; anyone with a torque wrench can use this method, but a lot of engineers are wary of it because of the variability of factors that influence the load. Friction is the main variable; it changes with materials, surface finish, lubricant, and is often worryingly inconsistent, even when we believe we have good control over all of the above. The application of lubricant or thread-locking compounds is a manual process, and variability is unavoidable. Therefore, even using the same tool, the operator is a variable.

The turn-of-nut method also has a couple of limitations. There is no torque developed until the nut is in contact with its adjacent joint member (ignoring 'prevailing torque' fasteners) and when we do have contact, there is a very non-linear portion of the angle-load curve. The first limitation can very easily be overcome, but the second cannot be avoided, as the turn-of-nut method is not accurate enough without detailed knowledge of the non-linear portion of the angle-load curve. Given that the non-linear behaviour is due in part to geometric inconsistencies, we cannot be sure we properly understand how the joint behaves in all circumstances.

All is not lost though. We can combine the torque tensioning and turn-of-nut methods to arrive at a more reliable process for pre-loading fasteners. In combining the two methods, we initially use torque control to take us just beyond the non-linear portion of the angle-load curve and onto the elastic part of the curve. The load induced in the fastener during the initial torque tensioning is modest in comparison to the final load that will be developed, and because the load is low, any effect from geometric inconsistencies or friction in calculating the load at the end of the torque stage is small enough not to cause too many worries.

Hereafter, we return to our turn-of-nut method. We are in the linear elastic portion of the angle-load graph for which we have measured or calculated data, and providing we can accurately measure the angle through which the nut is turned, we can be pretty sure we have got as close to our desired pre-load as it is possible to be using conventional tools.

Fig. 1 illustrates the method. Let us say that we want to induce a load of 25,000 N in the fastener. The linear elastic portion of the graph extends to a point where the load is 5000 N. We select a torque to take us just beyond this point, say 6000 N. For torque tensioning, ±20% accuracy might be reasonable, so the load could be 4800-7200 N. If we then tighten the bolt by 95º, the load should be 23,700-26,200 N. This is a range of 5.2% below target to 4.8% above.

For normal toque tensioning, we might expect a range of 20,000-30,000 N. If we can gain an understanding of the non-linear behaviour of the joint and the gradient of the elastic portion of the graph, we can predict bolt pre-load with a high level of accuracy.

Fig. 1 - Load-angle graph, showing non-linear and linear elastic behaviour

Written by Wayne Ward

Given that we know the pitch of the bolt, could we not simply arrive at an angle through which we turn the fastener (nut or bolt)? If we assumed the joint (clamped) members to be infinitely stiff, we could simply work back from our knowledge of the axial bolt stiffness and pitch to correlate tightening angle and force.

If the fastener has a stiffness k and pitch p, then for a single complete turn of the fastener head, we assume that the fastener has stretched by a distance p, and that the load is kp. For example, if the fastener stiffness is 50,000 N/mm and the pitch is 1.25 mm, then a single turn should increase the load by 62,500 N, if we assume that the joint is infinitely stiff.

Of course, we have immediately raised the first difficulty - the joint members are certainly not infinitely stiff. The finite stiffness of the clamped members must be taken into account, and if we imagine that the clamped parts are made of rubber, several turns of the fastener described above will develop very little load. Unless the geometry of the joint materials is extremely simple, the calculation of their stiffness is not a trivial matter, and the options for engineers without access to finite element analysis (FEA) software packages are either to turn to physical testing or to calculate the stiffness of the joint members based on assumptions and physical measurements.

The calculation of load is made more complicated by the finite stiffness of the clamped joint members, but there are other factors at play too. If the nut is not in contact with the washer when the nut is turned, no tensile load will be developed in the fastener, even if there is a resisting torque (for example from a locking mechanism). That much is obvious, so we may choose not to start measuring the angle of turn until the nut is in contact with the washer.

However, even with the knowledge of accurate fastener and joint stiffness, that is not going to give us the correct relationship between angle of turn and pre-load. The reason is that there is a portion of the tightening event during which the behaviour of the joint is non-linear. Until a certain load is reached, the gradient of the load/angle curve is not constant, and increases until it is equal to the gradient of the linear portion of the curve. If we assume that the load/angle relationship is linear, we stand to have significant differences between the actual fastener load and that desired, especially in cases where the design load is low and the non-linear portion of the chart takes up a significant part of the design load.

Fig. 1 - Knowledge of fastener stiffness, thread pitch and tightening angle won't guarantee accurate bolt pre-load

Written by Wayne Ward

Of course, in optimising the design of any given fastener, other factors such as the design of the joint components and materials selection play an important part, but one of the main causes of fastener failure, or the failure of the fastener/joint system to do its job properly, is poor selection and application of pre-load.

What might surprise you is the fact that the method by which the pre-load is applied can affect the design and mass of the fastener, because the differing methods by which the load is applied have varying levels of uncertainty associated with them.

There are two basic steps for the design engineer to consider - calculation of required pre-load, and determination of how the pre-load is applied. For this article, we will consider the application of pre-load by torque tightening alone. The relationship between force and torque is:

Where:
P is the pitch
µp is the coefficient of friction acting at the pitch diameter of the threads
dp is the pitch diameter of the thread
µw is the coefficient of friction acting at the nut-washer or bolt underhead-washer contact
dw is the mean effective diameter of the nut-washer interface

The relationship can be found in a number of books, notably the technical fastener manuals by Bickford*.

While the geometric elements of the equation are simple to measure, read from a drawing or table or calculate, there is considerable uncertainty in the coefficient of friction, and there is also inaccuracy in the application of torque using a torque wrench.

There are tables that give coefficients of friction for various combinations of materials, but roughness and choice of lubricant play a significant role which such tables do not account for. The number of times the joint has been tightened also has an effect. Some lubricants have been developed that are specifically aimed at providing a consistent relationship between torque and pre-load in fasteners.

Bickford estimates significant scatter of pre-load for a given tightening torque: ±30% of design pre-load for a run-of-the mill hand torque wrench and unlubricated fasteners, but this improves to only ±20% when using high-quality tools and lubricated fasteners. Fortunately, with the availability of specific fastener lubricants, as well as better tools and more consistent fasteners than were used for Bickford's experiments, we can expect to improve on these figures. So, why do we continue to use torque tightening as the most common method of pre-loading bolted joints? The answer lies in the simplicity of the method, the relatively straightforward calculations and the availability of good-quality torque wrenches.

* Bickford, J.H., "An Introduction to the Design and Behavior of Bolted Joints", Marcel-Dekker, 1981, ISBN 0-8247-1508-X

Fig. 1 - Good-quality torque wrenches are essential, as is regular calibration, if pre-load scatter is to be minimised

Written by Wayne Ward

Beyond the choice of materials, there are also a number of fastener coatings that are used for motorsport applications. These can range from those that are simply to prevent or delay the onset of corrosion when parts are stored, through to those that seek to change the tribological behaviour of the material.

We are probably all familiar with seeing black steel fasteners, with cap head screws being popularly used in most branches of motorsport. The black surface treatment does not alter the part dimensionally, and retains oil on the surface of the fastener, preventing corrosion. Also used for corrosion prevention are a number of metallic plating processes, notably zinc, chromium and cadmium. There are real dangers in the electroplating of fasteners, as mentioned in the recent article in Race Engine Technology (issue 59). Hydrogen embrittlement can lead to sudden failure of fasteners, and care needs to be taken to avoid this.

Where its use is allowed, titanium is favoured for its low mass and low elastic modulus. The matter of mass is simple, while the reasons why low elastic modulus might be favoured have been covered in various RET-Monitor articles and in Race Engine Technology magazine (see issues 41 and 59). The use of titanium is outlawed for use in 'static' fastener applications in a Formula One engine, but is strangely OK for use in Formula One chassis construction. Perversely, there is no such ban on the use of titanium fasteners in many lower-budget race series.

Titanium is famous for its tendency to gall at low levels of applied surface stress, even under very low sliding velocities and most especially against other titanium components. It is common therefore to find titanium fasteners that have their threads coated. Coatings involving lubricious compounds such as molybdenum or tungsten disulphide are often used and may be found to be combined with other surface treatments such as anodising. Other coatings developed for titanium fasteners, for military use, involve the deposition of aluminium onto the threaded portions of fasteners, then applying a further chromate conversion process and lubricating compound. The paper by Zurko* describes such a process.

The coating of fasteners with silver is common for motorsport use, especially where they are used in high-temperature applications such as exhausts or turbochargers. Silver has a low shear strength, making it a good choice as a solid lubricant, and it also is very resistant to oxidation. High-strength silver-plated steel nuts are commercially available in both metric and imperial thread forms.

Reference
* Zurko, M.J., "Evaluation of Sermetal W Coatings for Fasteners", Naval Air Development Center Report NADC-75121-30, 1975

Fig. 1 - Silver plating of nuts is just one of a wide range of coatings used on motorsport fasteners

Written by Wayne Ward

Other companies are much happier to supply plated fasteners, however, having taken every precaution to minimise the chances of embrittlement before, during and after the actual electroplating process. Although we see fewer zinc-plated fasteners being used in high-strength materials, cadmium plating is still a popular process - 'NAS' airframe bolts are an example of a cadmium plated fastener - and silver plating is often used to prevent seizure of materials prone to galling or in situations where conditions are too harsh for lubricants to continue to work over long periods.

The reason for the strong opinions held by some is due to a phenomenon called hydrogen embrittlement, which leads to a drastic reduction in fatigue life and static strength. While there are a number of precautions that may be taken to lower the risk of damage from hydrogen embrittlement, ISO 4042, an international standard relating to electroplating of fasteners, states, "Complete elimination of hydrogen embrittlement cannot be assured."

During the electroplating process, and commonly also in any earlier cleaning and descaling processes, hydrogen is evolved at the surface of the components and, particularly in the case of steels, is absorbed by the surface of the metal. Higher strength steels are particularly sensitive to hydrogen embrittlement.

In order to reduce the risks, mechanical cleaning and descaling processes are recommended, and acid cleaning is said to present a greater risk of damage than alkali processes. In an acid, iron reacts to form a salt, with the result that hydrogen (H+) ions thus liberated combine with electrons to form hydrogen gas.

Any electrochemical cleaning processes used should be carefully selected by the plater; cathodic process liberate hydrogen, whereas anodic processes do not.

Following the plating process itself, it is widely agreed that a baking process, carried out soon after plating, can be an effective way to reverse much of the damage that occurs during plating. This is done at relatively low temperature (around 200 C) and there are guidelines for how long it process should be carried out for (1) according to the strength of the steel, with higher-strength alloys requiring longer treatments. For steels with a tensile strength of 1700-1800 MPa (247-261 ksi), the minimum recommended time for a post-plating bake is 22 hours.

We should not be under the impression though that only steel fasteners suffer from this problem, although they are probably the most likely candidates for plating processes as far as fasteners are concerned. Titanium, nickel and aluminium materials have been reported by Dini (2) as suffering from this phenomenon. However, some of the more exotic fastener materials are known not to suffer from hydrogen embrittlement.

It should also be noted that it is not only electroplating processes that cause steel fasteners to suffer in this way. Electroless plating processes, such as electroless nickel, also produce hydrogen embrittlement, as can phosphate coatings in a more limited way.

1. Raymond, L., (editor), "Hydrogen Embrittlement: Prevention and Control", ASTM STP 962, 1988

2. Dini, J.W., "Electrodeposition", published by William Andrew, 1993, ISBN 0-8155-1320-8

Fig. 1 - In solving a corrosion problem in plating fasteners, we should be careful not to damage them during the plating process

Written by Wayne Ward

Materials specially selected and processed to have high fatigue strengths are combined with design features that minimise stress concentration. Particular attention is paid to the underhead region of headed fasteners, and any transition from one diameter to another is generally achieved with a large radius. Also, very long fasteners may have supporting features along the shank to mitigate the effects of vibration.

One technique you will very rarely find used with any success on a highly stressed fastener is a machine-cut thread. While such machined threads are fine in some static applications, they are generally shunned where high cyclic stresses are found. Thread rolling is the preferred technique to produce fatigue-resistant male threads.

The process of thread rolling is carried out on a portion of plain-diameter bar that is carefully chosen to produce the correct result after having been rolled. The pre-rolling diameter is chosen according not only to the thread being formed, but also the material being rolled. Rolling is a cold-work process during which the workpiece is plastically deformed to produce an accurate thread.

While there should be very little geometric difference between a machined and a rolled thread, the difference in performance comes from the fact that the rolled threads have been produced by cold-working; the result of this is to leave the thread surfaces, including the highly stressed thread root, in a state of residual compression. This residual compressive stress serves to reduce the stresses induced by the working load on the bolt, producing the same effect as having a lower applied stress in a machined thread. We can therefore see that the effect of introducing the residual compressive stress due to rolling is to increase the fatigue strength of the fastener.

While we now have a better understanding of the effect of thread rolling and methods of predicting and quantifying the residual stresses thus produced, this has not always been the case. In their book of the early 1950s, Lipson et al* ascribe stress concentration factors to screw threads, and these are different for machine-cut and rolled threads, with a significantly higher factor being used for machine-cut threads.

Thread rolling is a specialist process; while it offers some very obvious benefits, it can also be done badly if you don't choose the right supplier. If you are designing and buying bespoke fasteners for the first time, you should take care to select a manufacturer with a good reputation and who can offer you advice on the design of your fastener to suit the rolling process.

*Lipson, C., Noll, G.C., and Clock, L.S., "Strength of Manufactured Parts", McGraw Hill, 1950

Fig. 1 - These are typical thread rolls used for producing high-quality male threads

Written by Wayne Ward

In many circumstances, socket-head screws also have advantages in terms of tool access. As engines and components become ever more compact, the use of smaller screws becomes more prevalent. However, in the case of small screws - and button-head screws in particular - the use of standard socket-head screws will often cause technicians who have to deal with these components to moan.

The reason is that the underhead friction is high, and the socket is small and often less than perfectly formed. Given that female hexagon drive features of all types drive on the corners rather than the flanks of the socket (there are special tools for male hexagon drives that drive on the flanks of the bolt heads), there is a tendency for small screws to 'round' - the annoying phenomenon where the socket deforms to leave a round hole, inside which the tool cannot drive.

At this point the technician has to resort to one of a number of time-consuming 'non-preferred tactics' for removing the fastener, which often involve a hammer and a punch or a chisel. Worn or undersized tools, slightly oversized sockets, and weak fastener material all serve to exacerbate this problem (always be aware of the grade of fastener that you are being supplied with).

So, what can we do to minimise the chances of rounding? Well, if we are already using a well made and strong fastener, then we need to look at an alternative to the hexagon socket drive. There are a number of designs that drive using the flanks of the tool, the most commonly known of which is called 'Torx'.

The general form of female flank drives is to shape the recess in the form of a multi-pointed star. While it is unlikely that the socket and tool are so perfectly formed that there is immediate full surface contact on all flanks, there is relatively little deformation required such that the drive is essentially one of face-to-face contact on each flank.

On a hexagon drive, the reaction of the 'points' of the tool are at 60º to the tangential drive direction, - that is, the reaction is more radial than tangential - and therefore there is a strong tendency for the screw material to be deformed outwards.

Flank drives have a much smaller angle between the tool reaction and the tangential direction, and therefore a relatively lower tendency to force material radially outwards. Also, there is a greater amount of material between the circles inscribing and circumscribing the socket feature for flank drives than hexagon sockets. Torx or similar pan-head screws seem generally preferred in small sizes where rounding is a common problem for these reasons.

Fig. 1 - Flank-drive socket-head screws offer advantages over conventional hexagon socket screws

Written by Wayne Ward

There have been several attempts to manufacture nuts with a thread form that distributes the load evenly between threads. You will recall from the previous article that use of standard nuts places the vast majority of the load on the first few engaged threads.

The Spiralock thread form has been shown to improve load distribution, and nuts with this thread are commercially available in a number of forms, all of which are compatible with standard male fasteners. It is also possible to buy taps in various configurations, including cold-forming taps (which form threads by material displacement rather than removal). It is therefore possible to design and manufacture small batches of bespoke fasteners that use this thread form.

The action of the thread relies on there being line contact between the outside diameter of the male thread and a specially designed portion of the female thread. Contact pressures are therefore high (standard threads assume face contact) and some permanent deformation of the female thread is expected. There is also a change in the torque-tension relationship owing to two factors - effective contact angle and contact diameter. The effective contact diameter moves from the pitch diameter to the outside diameter of the male thread. A Spiralock thread is said typically to require 10-20% extra torque in comparison with a standard nut to attain a given level of tension in the fastener.

Equa-Stress threads were a patent of ESNA, now part of SPS. The modern equivalent is called Double Durability, and a number of different pattern nuts are available, although little literature exists to explain where the improvement in fatigue life comes from, and whether there is any effect on the torque-tension relationship.

Bolted joints using differing flank angles on the male and female parts have proved effective in increasing the fatigue life of bolted joints. A study in the former USSR some years ago showed a 26% increased in endurance limit for standard bolts engaged with nuts having a smaller included thread angle.

Asymmetric threads have also proven positive in this regard, with both nuts and bolts having asymmetric threads showing improved fatigue limit when engaged with standard mating fasteners. However, care needs to be taken in designing and using such components, as changing the angle in the wrong direction will lower the endurance limit.

Over-pitching of nuts - that is, increasing the pitch slightly compared to the mating component - has also proven successful in improving fatigue properties by distributing the load more evenly over the engaged threads. The same effect can be achieved by de-pitching of bolts. The philosophy behind this is simple. For standard fasteners, as the joint is loaded, the bolt is stretched and the nut is compressed, giving a pitch error that causes a poor distribution of load. In over-pitching the nut thread, the thread pitches can be engineered to match in the loaded condition. The optimal amount of over-pitching (or de-pitching if applied to a bolt) depends on the material properties, bolt dimensions and thread form.

Fig. 1 - The internal thread form of a nut, in conjunction with a mating part having a conventional thread, can improve endurance limit markedly

Written by Wayne Ward

There is much published material on the distribution of loads in the threads of nuts, and this has been referred to very briefly in a previous RET Monitor article, and also in the pages of Race Engine Technology magazine in a Focus article on fasteners.

The rest of this article will look at some aspects of nut design and material selection that have been studied by various engineers in the former USSR when working on fundamental research into the strength of threaded connections.

If we consider a stud and nut, the load path through the stud and nut requires that the lines of axial force are reacted at the thread flanks, and then the resultant compression and bending loads are returned immediately to the nut face. Improvements in stud fatigue life were noted by using two design methods - machining a groove into the washer face of the nut, and by employing a 'stepped nut'. Both methods have similar effects, and for the same two reasons.

First, the initial engaged threads are less stiff than those above. Given that the initial engaged threads commonly carry vastly more load than those further from the joint face, this promotes a more even distribution of load than in a conventional nut. Second, in each case, the load path is not turned so severely at the first thread, thus giving a lesser stress concentration factor. By employing a nut with a very narrow, deep groove in the face, the fatigue strength was increased by 30%, and improvements of between 18% and 30% were measured in the case where a stepped nut was used.

The manufacture of a grooved nut with a narrow, deep groove would not be easy for many of the small nuts generally used in racing, and the use of a stepped nut would not be practical in some circumstances owing to the increased space required.

In a separate experiment, two changes to the nut material were also found to have a significant effect on the fatigue strength of the stud. The first was to take the existing nut material and decrease its strength, in this case by using a normalising heat treatment. The second method was to use a material with a lower modulus than the existing steel nut. Compared to a geometrically identical nut, the normalised nut increased the stud fatigue strength by almost 6%. The use of a lower modulus nut material increased the fatigue strength of the stud by 40%.

Reference: Yakushev, A.I., "Effect of Manufacturing Technology and Basic Thread Parameters on the Strength of Threaded Connexions", Pergamon, 1964

Fig. 1 - A stepped nut, of the kind that gave a 30% improvement over a standard nut in a study by Birger in the former USSR. The flange is fitted uppermost

Written by Wayne Ward

The torque-tension relationship relies on a number of geometric parameters that are often under our control, or are things that we can measure directly. One thing we need to ensure, where we use a washer under the head of a bolt or under the seating face of a nut, is that the washer remains static while we tighten the bolt. Otherwise, we are having to overcome friction at an interface with a different dimension from those we have used in our calculation of washer-face friction. This gives another layer of uncertainty to our calculation.

The washer dimensions are generally larger than those of the nut face, and the contribution of friction from the interface between the washer and whatever it sits on will probably be larger than you were expecting. Washer rotation should only be prominent when the friction coefficient between the washer underside and the adjacent part is lower than that between the nut face and the washer. This may happen if the casting is a little oily and the nut and top of the washer are dry. Of course, this isn't the usual situation, but it may happen in less than ideal circumstances - late night, rushed build and so on.

Another good reason for preventing any rotation of the washer, particularly in joints with high contact pressure, is to prevent the washer from damaging the casting if it rotates. There are many ways of doing this, but the most popular is to cause some 'damage' to the underside of the washer. Centre-punching the washer to produce dimples with slightly raised lips means that, on tightening, there is a local high-pressure region and the casting material will 'flow' into the dimple, thereby locking the washer in place. Doing a similar operation with a chisel mark across the washer face will have a similar effect.

Some people produce machined washers with a narrow machined groove across them for the same reason. Any feature that gives a local high pressure will have the same effect. Be careful though to ensure that these anti-rotation features are on the underside of the washer when it is installed.

Other schemes are often more complex, and involve non-round washers in specially shaped pockets to restrain rotation, or other similarly involved practices. Many of these alternatives to the centre-punched or chisel-marked washers are impractical from the point of view of packaging them within an engine. The simple methods don't require any extra space or special machining methods.

The centre- or chisel-punching methods are only really suitable for thick washers, as they cause excessive deformation of a thin washer, and this is likely to have the wrong effect and run the risk of sitting the nut face on a local high point(s), with little lubrication and therefore increased friction.

For those who feel that punching or similarly marking their washers is the work of heathens, then machining a groove or centre drilling the face of the washer is a good alternative that has less chance of distorting the washer.

Fig. 1 - Washers, if they rotate during tightening, affect the relationship between tightening torque and fastener tension

Written by Wayne Ward

For critical applications, we would ideally like to measure the tension in a fastener directly, but this is rarely possible. Our next best method of gauging the tension in a fastener immediately after tightening is to measure its length and compare this to its free length. Given some accurate information regarding the geometry of the fastener and the properties of the material from which it is made, we can calculate the load in the fastener. Unfortunately, this is possible only for fasteners where we can access both ends to make measurements using a 'stretch gauge' or in fasteners large enough to incorporate measurement features.

For many engine suppliers, the best available method is to specify a tightening torque and calculate the load based on this. The relationship between torque and tension depends on a number of parameters, including the thread flank angle, its pitch, pitch diameter, dimensions of the mating face of the washer/nut/mating face of the bolt as applicable, and the friction at any interfaces. The coefficient of friction is important because the higher it is, the less tension we will achieve in a given bolt for a certain level of torque.

One reason why torque specification for bolt tightening is not the first preference of many engineers is that there is large uncertainty in the level of friction. There are a number of sources of friction coefficients, but in practice the actual level of friction can depend on many factors, including the cleanliness of the components, their surface finish, whether the joint has been 'preconditioned' by previous use, and the type and amount of lubricant used.

The friction coefficient also has an effect on the level of stress within the bolt, both during and after tightening. As we know, when we tighten a fastener we induce a level of twist in the shank and threads, which is proportional to the torque used. From our physics and engineering classes we know that the angle of twist is proportional to the applied torque. The shear stress due to torsion is also proportional to the applied torque. The torque involved in twisting the bolt is proportional to the coefficient of friction, and therefore we can see that the shear stress - and the combined stress due to shear and tensile stresses - is affected by friction. It is known that a proportion of the shear stress relaxes in the period immediately after tightening, but we still have to consider its effect in yielding the fastener during tightening. There are special methods which can preload fasteners without inducing any torsional shear stresses, but these are generally restricted to larger fasteners than we would find on a racing engine.

Minimising the level of friction and being able to control its consistency is important if we want to have faith that we have achieved the desired level of tension in the fastener. Special lubricants have been developed for this purpose and have been discussed in a previous article.

Fig. 1 - Thread and underhead friction has a critical effect on the torque-tension relationship in fasteners, and also affects the maximum stresses in the fastener (Courtesy of ARP)

Written by Wayne Ward

Have you ever wondered why this is so? It is quite often the case that studs are installed into a casting made of a low-density material such as aluminium or magnesium, and it is the relatively low strength of the metal into which the stud is inserted that gives rise to the requirement for a large, coarse thread. If we have a strong stud, and a weak casting, it is easy to imagine the mode of failure if too great a pre-load is applied: the female thread in the casting will 'strip' and be pulled from the casting. This is a common failure, and results from shearing of the casting at the major diameter of the stud.

If we consider the total area in shear, it will be proportional to the major diameter of the stud multiplied by the length of engagement. It also depends on other factors such as the tolerances that have been applied to the female thread, but in all cases the basics should hold true - the shear area increases in proportion to diameter and engaged length.

The rule of thumb concerning stud shear load being proportional to stud major diameter (for a given thread pitch) holds true. However, practical experience will show that the load to cause failure by thread shearing is almost independent of engaged thread length beyond a given length, and this length is surprisingly short.

The reason for this can be found in a large number of engineering textbooks, especially those concerned with threaded fasteners. The distribution of load along the engaged length of a stud (or nut) is very far from being even, especially where the modulus of the casting and stud are equal.

For more than 100 years, engineers have been aware of the fact that the engaged threads closest to the applied tensile load take the vast proportion of the force. Work published in Russia in the early 20th century showed that for a steel nut and bolt with ten engaged threads, 34% of the applied load is taken by the first thread, and 85% on the first four threads.

The load distribution changes little for longer threads, and is not much affected by thread pitch. The shear area can therefore be said to be proportional to pitch rather than engaged length. As the first thread distorts under load, so more load is transferred to the second and subsequent threads, but it is quite common to fail the threads closest to the applied load before any real damage has occurred to the lower threads.

The uneven distribution of load in the engaged thread can be influenced by changing the combination of materials, and using various design features. However, the rule that the load to cause failure increases with the strength of the casting material, the major diameter of the stud thread and the pitch of the stud, is worth committing to memory.

Fig. 1 - This part is typical of many high-quality bespoke studs, having a large-diameter coarse thread on the installed end (Courtesy of Blanc Aero)

Written by Wayne Ward

A European company has developed a new form of composite fastener that is a pure polymer-matrix component. Used in a number of applications initially, the company isn't actively marketing to the motorsports or other industries, choosing instead to focus on its largest market, the medical industry.

There are a number of applications where these screws would be of use in motorsports powertrains. In quasi-static low stress applications where a steel, titanium or aluminium screw might be over-engineered, a high-strength polymer fastener might be a good alternative.

The properties of polymers make them useful for many applications. In general, polymers are poor conductors, both of heat and electricity, and this makes them an attractive prospect for high-voltage systems and where we would like to restrict the flow of heat. In Formula One, the engine rules state, "All threaded fasteners must be manufactured from an alloy based on cobalt, iron or nickel"; however, in other race formulae, such polymer fasteners could find use in engines.

The problem with pure polymer fasteners is that they are very weak compared with a metallic fastener, so the concept of reinforcing the part with a fibre has a lot of technical merit. The company mentioned above has a manufacturing method by which the core of the fastener is reinforced longitudinally, and the threads are also reinforced by fibres that run helically inside the threads. This provides a much stiffer and stronger structure than would be provided by a pure polymer component.

The matrix material used is a high-strength, semi-crystalline polymer that has good properties at high temperature - the parts can be used at close to 200 C, making them suitable for many applications for design engineers. The fibre used is carbon, although there is no particular reason why it can't be replaced by other fibres. The fact that the fibres are surrounded entirely by the matrix and are not present at the contacting surfaces (underhead and threads) of the component means the parts maintain their electrically non-conductive properties, despite having a conductive reinforcement.

With many polymer fasteners, the options for manufacture are to machine from solid or mould the finished item directly. In terms of a conventionally moulded part, the options for reinforcement are to mould using a material with short fibres that can be used with injection-moulding equipment. However, the advantage of these particular fasteners in terms of mechanical properties, as the accompanying diagram shows, is that they have continuous long-fibre reinforcement running through the length of the part. While the volume fraction of reinforcement is less than we would find in a short-fibre injection-moulded part, the fibres are oriented in a much more favourable direction for load transfer.

Fig. 1 - The diagram shows the orientation of fibres in a typical screw, along with some of the component forms available

Written by Wayne Ward

There are many areas in a Formula One engine, for example, where composites are expressly forbidden, but in production-based series, people are free to use composite components much more freely and so they have developed techniques to make their use economical. As with any new technology, the early adopters have to spend large sums for questionable gains, but as the new technique or product is taken up by a wider audience, the price drops and it becomes more affordable.

One British company looked into producing fasteners from composite materials a few years ago, initially for the Formula One market. As we might expect, this new technology was crushed from on high, but it is worth examining the technical pros and cons of these products.

In terms of looking at the physical parts, the threaded portion of the fastener was, as we might reasonably predict, made from metal, while the shank of the fastener was made from a composite. One of the composite fasteners is shown in the photo here. The term 'composite' can be applied to this component on two levels - not only is the shank made from a composite material itself (carbon-fibre reinforced plastic) but the fastener as a whole is made from more than one material.

One technical difficulty is in attaching the metallic threaded portions of the fastener to the composite shank. Such fasteners are expected to bear heavy loads and to cope, in many cases, with a significant cyclic load as well.

As can be seen from the photo, the plain shank of the fastener is of a smaller diameter than the composite, which can be seen at the end of the threaded portion. We can discern from this photo that the composite shank spreads to form a wedge within the threaded ends, thereby eliminating any adhesive bond to perform secondary functions such as preventing the shank from rotating, although it is likely that this is also not left to the integrity of the bond between the ends and the shank alone. In the case of the fastener shown, the ends are made from titanium, but the technique clearly allows the use of any material for this purpose.

So what advantages might a composite shank confer on a fastener? As we have seen from previous articles, it is likely that we want a combination of properties that include high strength and low modulus. The first dictates to a large extent the durability of the part, while the second helps to limit the proportion of the cyclic load in the joint which the fastener must be designed to cope with.

Looking at the properties for a typical fibre used in carbon-fibre reinforced polymer (CFRP) pre-preg (carbon fibres pre-impregnated with resin), which would be indicative of the material used for this application, we can see that a tensile strength of 4800 MPa (700 ksi) is typical, combined with a modulus of about 240 GPa (35 Msi). If we assume that the volume fraction of polymer used is sufficient to reduce the strength to 2400 MPa (50% volume fraction of fibre is typical for pultruded forms), then by a simple scaling calculation the modulus of the composite would be 120 GPa. Pultrusion is a similar process to extrusion, except that it is impossible to extrude, or push, fibres without them becoming distorted, so fibre-reinforced composites need to be pulled to form a continuous section such as a rod with fibres running lengthways.

So, for a strength of 2400 MPa, in excess of the typical strength levels achieved in high-specification metallic fasteners, we achieve a modulus comparable to that of titanium. The composite shank in this example can therefore have a 17% lower cross-sectional area, in a material with a density of about 2g/cm3, roughly 75% lower than that of steel.

There is a wide range of fibres that offer a combination of properties which would prove more favourable for a fastener than those presented here. The development of a fastener with a CFRP shank will surely not remain repressed for very long, even if motorsport doesn't take advantage of it.

Fig. 1 - Experimental Formula One stud with CFRP shank and Ti threads (Courtesy of Crompton Technology)

Written by Wayne Ward

Where a fastener is cyclically loaded, we need to consider the effects of fatigue, and in this regard we need to pay special attention to design features. Where a bespoke engine is concerned, or where we are looking to increase markedly the output of an existing engine, it is very likely that we will have to look beyond the realm of 'off-the-shelf' fasteners and towards something that is specifically engineered.

For popular conversions of production engines for motorsport, there are improved fasteners for critical applications such as main bearing caps and cylinder head studs, and it may be that these are sufficient for a great many engineers who are tuning production engines.

There will be many applications, however, when the fasteners available as standard from race specialists won't be up to the task. In this case, obtaining a component of sufficient durability might simply involve having the part made in a better grade of material.

In bespoke engines, where the engineering is better optimised, and space generally tighter, there are a lot of highly stressed special fasteners used throughout the engine, and almost certainly in critical areas such as the con rod bolts, main bearing cap studs and cylinder head bolts/studs. Common design features such as large blend radii and waisted shanks are used to improve fatigue life, in these cases by minimising the stress concentration factor and lowering the internal load coefficient respectively.

It is usually at the thread, however, where fastener problems are to be found, which may be due to reasons too numerous to list here. One action we can take though is to look to the geometry of the threadform itself for an improvement, and fortunately there are threadforms with a tightly controlled root radius in both imperial (unified) and metric dimensions.

These are known as 'J' threads, and you may see designations such as MJ or UNJF used to describe threads cut to this standard. They are generally compatible with standard threads of the same basic form, but have an increased radius at the thread root (minor diameter) compared with the standard type.

One point to be careful of is to check the specification of the nut or internal thread used with a 'J' specification thread, as it may not have a sufficiently large internal diameter to ensure clearance to the increased root radius. An increased root radius serves to decrease the stress concentration factor while slightly increasing the minor diameter, both of which serve to lower the critical stresses in the component.

Where a standard thread has a root radius which is a minimum of 0.125 times the thread pitch, a 'J' specification thread has a root radius controlled with the range of 0.1501 to 0.1804 times the thread pitch. Thus the minimum root radius of the MJ thread, for example, is 20% greater than the guaranteed minimum on a standard M profile thread.

Fig. 1 - Race engine fasteners such as these studs are often specified with 'J' form threads (Courtesy of T&K Precision)

Written by Wayne Ward

It is now pretty rare to find a race engine airbox, certainly on 'formula' cars, made from anything other than carbon-fibre reinforced polymer (CFRP) composites. It has also been used to good effect for plenums on turbocharged engines, on structural covers for race engines and in parts used specifically to increase the stiffness of the engine in order to provide a performance benefit to the vehicle as a whole. It is widely used for electrical boxes and cosmetic covers too.

One difficulty with using fibre-reinforced composites though is that they cannot be directly tapped in order to provide a strong thread. To provide a thread, we really need to provide some metal into which the thread is already cut, or in which is can be formed. There are a number of methods of doing this, some of which we will discuss briefly here.

Clearly it is possible, where access allows and using a conventional nut, to provide a metal thread, but this can be fiddly. A flanged nut can be bonded to the composite panel, and there are companies who make nuts that are specially designed with a large flange and features to improve the reliability of the bonded joint. These bonded fasteners can be used during the lay-up of the part if required.

The most basic way to provide a permanently fixed thread is to bond in a piece of metal into which a thread is formed. These 'hard points' can be made in a material that best suits the application, and have the further advantage that they can take whatever form is required. The chief disadvantage is that they are more costly than the proprietary parts discussed.

'Rivnuts', as the name suggests, are nuts that are riveted to sheet parts. They are basically a thin-walled tube, provided with a thread at one end and a flange at the other.

The tube is pushed through a clearance hole until the flange is against the composite panel, and then a special tool crushes the tube against the other side of the panel, thus restraining the rivnut. They are easy to use, but not all types leave the flange flush with the panel.

Anchor nuts are fasteners with a female thread and which are riveted to sheet metal or composite parts, and can be fixed or 'floating' types. The floating type is pictured here and, as the name suggests, the thread element has a certain amount of lateral movement and can accommodate inaccuracies in the hole positioning on the composite part or in the mating piece. They are also available with multiple nuts on a single plate, meaning that fewer individual parts are needed.

Fig. 1 - This floating anchor nut is typical of the type of fastener available for composite applications

Written by Wayne Ward

One of the more reliable ways in which we can tighten a fastener is to measure the extension or 'stretch' of a given bolt or stud, and for a con rod, this is indeed the way that the vast majority of engine builders will work. Knowing the load versus extension relationship of the bolt means we can specify an extension range to where the bolt must be tightened.

There are many situations, however, where the fastener can not have its extension measured and we therefore need some knowledge of the relationship between torque and tension. This relationship depends on a number of factors such as the diameter and pitch of the thread, the contact geometry of the fastener underhead and the level of friction in both the thread and underhead areas.

Where the physical geometry of the components is fixed, the friction can be affected by choices of materials and coatings, and also by the use of any thread lubricant. Having undertaken a programme of testing on bolt pre-load versus torque, I can say this relationship is not consistent, even with conditions as close to ideal as possible.

There is a great deal of scatter between the pre-loads achieved during each individual tightening of a fastener, and this is problematic. The consequences of insufficient pre-load can mean loss of performance, or indeed catastrophic failure and total loss of a very expensive engine. It is common practice for companies to use engine oil or 'moly' paste, 'copper-slip', EPL or one of the many alternative lubricants in critical applications, but there is little evidence that any of these is significantly better than any other in real terms.

A well-known fastener company has taken the step of developing a special thread lubricant to improve the torque-preload relationship and thereby provide us with a more reliable fastener. There are a number of ways by which we might judge this to be a success, and the two graphs show measured test data which we can use as criteria of success.

Fig. 1 shows the pre-load achieved by tightening the same bolt on a number of occasions; the torque specified was 120ft-lb. The red trace corresponds to the new lubricant, and we can see that it shows excellent consistency compared to the other data.

As many of you will be aware, a joint generally takes a number of tightening cycles to 'bed-in' and this phase is shown in the blue, yellow and green traces as a rising preload as the joint is subjected to repeated assembly and disassembly. This phase of 'conditioning' the joint is eliminated with the new lubricant.

Fig. 2 shows the effect of tightening ten new fasteners, and is a good illustration of what we might do in tightening the cylinder head joint, or main bearing caps. Again, we see a surprising level of consistency here.

In the worst case, with oil - which is what many people use in this situation - we see that we may achieve only about 65% of the intended preload. The lower level of preload could be mitigated by specifying a higher torque, but this wouldn't change the poor consistency of the results. Furthermore, we would increase the chance of damage to the fastener owing to higher shear stresses due to the higher torque.

The new lubricant, called Ultra-Torque, is made by racing fastener company ARP (Automotive Racing Products), and I asked its director of specialty products Chris Brown to explain the advantage of Ultra-Toque for new fasteners.

"Oil, moly and CMD/EPL do not meet the test objectives," he says. "All three lubricants have a large amount of preload scatter requiring six to eight cycles before achieving the recommended preload. ARP Ultra-Torque clearly allows the fastener to reach the optimum preload level on the first cycle, and becomes repeatable within 5% on all remaining cycles.

"Consistency on a Cycle Test will pay major dividends on align honing and cylinder honing operations, saving the machinist time and money. The less preload scatter you have, the more repeatable your cylinder and housing bores will be on assembly, reassembly, and mock-up procedures," he says.

Fig. 1 - The pre-load achieved by tightening the same bolt on a number of occasions
Fig. 2 - The effect of tightening ten new fasteners

Written by Wayne Ward

The 'problem' with solid inserts is that they can (and generally do) bias the distribution of load towards the first engaged threads, leading to a higher stress concentration factor for the male threaded part. This is because, almost without exception, they are installed in parts made from materials with a low elastic modulus.

But since these will be used in most instances as a repair scheme, should we consider this a problem? In most cases the answer will be no. Clearly it is the female thread that is the weaker of the two in these cases, and even though we will cause a worse load distribution in the male threaded component, it is often preferable to scrapping an otherwise serviceable cylinder head, for example. It would be rare to find that the male threaded component would have such a marginal factor of safety in these circumstances, such that fitting an insert would transfer failure to the screw, bolt or spark plug for example.

Solid inserts are generally equipped with a self-locking feature which means that replacing them can be difficult. The popular 'Timesert' solid thread insert is one such component, and the locking mechanism relies on deforming the bottom thread of the insert radially outwards, locking it in place. The insert has a flange at the outer end, and the installation tool flares the lower part of the insert once the flange has 'bottomed'. These inserts are a popular repair for stripped spark-plug threads, and special kits are sold for this purpose for some popular models of cars.

There is another kind of insert which is locked in place by a more aggressive mechanical feature. 'Keenserts' are solid inserts with square-section locking 'keys' cleverly located in the sides of the inserts. Once the insert is installed to the correct depth, the keys, which are designed so that they have a cutting action as they are installed, are hammered into place, firmly locking the insert in position. The Keensert is primarily designed for insertion into low-strength materials owing to the requirement to force the 'keys' into place.

As repair schemes, both types work on the principles of first removing the damaged female thread and then enlarging it. The enlargement means that the thread in the casting in undamaged material and the area of the thread in shear is increased, thereby increasing the load capacity of the female thread in the casting. The bolt is installed in a female thread of higher strength material and therefore, even though we have replaced one female thread with two, the chances of either failing are very low.

Fig. 1 - The picture shows a 'timesert' installed in an aluminium casting

Written by Wayne Ward

Previous articles that I have written on the subject of fasteners have talked of the influence that the effective stiffness of the female thread has on the distribution of load within the fastener. There are plenty of references to load distribution in threaded fasteners in textbooks and the reader would be well advised to make himself / herself familiar with this, especially if you have the task of designing some or all of the components which make up a bolted joint. To recap, the majority of the load taken by the female thread is concentrated on the first few engaged threads, irrespective of the total length of thread engagement. The effect of this is to increase the stress concentration on the bolt or stud.

Specifically, we should understand that a more flexible female thread offers a more even load distribution. This can be achieved through using a female thread with a lower modulus of elasticity (the measure of the stiffness of a material) than the male threaded component, or by other design features applied to the female thread. However, it is quite often impossible to change the material of the female thread for various reasons.

The wire insert has an inherent flexibility which can, in some circumstances, serve to mitigate the stress concentration due to load distribution in the bolted joint, thus improving the fatigue life of the components in question. Hopefully you won't find yourself in the situation where you have fatigue problems concerning bolts, studs or expensive castings, but the wire insert is a solution to bear in mind for certain applications. However, if you are thinking of using wire inserts for this reason, you should be sure that it will achieve the intended effect, i.e. to make the female thread more flexible. There are situations where taking the step of fitting a wire thread insert could actually make things worse from this point of view. An example of this might be where the female thread is already suitably flexible and that it is some aspect of the bolt design or its material mean that its fatigue strength is marginal. Here, the addition of a wire insert may serve to stiffen the female thread and increase the stress concentration, thus hastening the fatigue failure.

Moving away from the idea of using a wire insert to change fatigue behaviour, a further advantage of the wire insert is the ability to 'renew' the female thread as required in case of damage. There are several ways to remove a wire thread insert, including special tools specifically designed for the task, one of which is shown in the accompanying picture. Providing that the thread into which the insert is fitted remains in good condition, you can fit a new wire insert, thus renewing the thread.

Fig. 1 - Special tool for removing wire thread inserts

Written by Wayne Ward

Each of the low density materials has, in general, an Achilles' heel where female threads are concerned and that is the low strength of the material, especially where repeated disassembly is a factor. Small diameter threads are often a problem in these materials and sometimes, after little use, the threads can be in a state of distress. Sometimes the favoured repair is to move, where possible to a larger thread, but in a lot of cases, we will reach for the thread repair kit, often referred to as the 'helicoil' kit.

Helicoil is a trade name and is often used to collectively describe any wound wire thread inserts, although they may not actually be 'helicoils'. Fitting a helicoil is a multi-stage process requiring special tooling and some care. Initially we must remove the damaged thread, and this is done by the simple process of drilling. We then take a special tap, which is of a larger size than the thread to be repaired, but of the same pitch. After tapping the hole to the correct depth, the wound wire insert is fitted using a special tool, normally supplied by the manufacturer of the insert. Generally, the inserts are wound into place using a 'tang' on the end of the insert, and this must be broken off before the insert is used with a bolt / screw.

Wire inserts are generally available in lengths which are multiples of the nominal screw diameter, and also in a range of materials; stainless steel is the default material, but others are available including bronze and titanium. A range of surface finishes are also available including silver plating.

There are different types of wire inserts and some incorporate a self-locking feature which acts much as a prevailing-torque nut does. One coil is deliberately deformed to provide an increased torque on the fastener. In areas where a loose screw may be a risk, as mentioned in the previous article on locking nuts, these screw-locking inserts are often used to good effect.

Whilst many consider wound wire inserts to be a repair scheme only, it is common to find them specified in new components, either to increase longevity or to increase the strength of the thread, allowing a shallower thread to withstand the design pre-load without damage to the female thread. We shall look at other forms of thread inserts in a future article.

One point to bear in mind with wire inserts, or any kind of ferrous insert in aluminium parts, is that they must be installed after anodising, if this process is specified. It leads to severe damage of both the insert and the aluminium part during the anodising process.

Fig. 1 - Wound wire insert showing drive tang.

Written by Wayne Ward

There is an advantage to designing and using a bespoke nut which many do not take advantage of, and that is that the nut can be designed such that it can improve the fatigue life of the bolt or stud to which it is attached. Where there is a problem with a bolt of a stud which causes a fatigue failure at the face of the nut, this is because of the stress concentration at the first engaged thread. In many cases the engineer will look to strengthen the stud or bolt which is a sensible measure and by doing so he may increase the fatigue strength of the component quite markedly. However, it is possible to make an even greater improvement in the fatigue strength of the bolt by a number of simple changes to the design of the nut, and the material from which it is made.

It is a widely accepted fact that the distribution of load on the engaged threads of a bolt is far from uniform. In the recent Race Engine Technology Focus article on the subject of fasteners, the author wrote "When we come to calculate fatigue life, we see from literature that the load borne by the threads is not at all even. Zhukovski first proposed the now accepted hyperbolic load distribution as early as 1902. The effect is that, for a steel nut and bolt, with 10 engaged threads, 34% of the load is borne by the first thread, and 85% is carried on the first four threads alone." From Zhukovski's work, we can imagine that there is a significant stress raiser at the first engaged thread due to the concentration of stress. We can markedly reduce this stress concentration factor by changing the material of the nut, the design of the nut, or both. In terms of materials, it is possible to improve the uniformity of stress distribution by using a nut of a lower elastic modulus than the bolt or stud. If we are using a bolt made of steel or a superalloy, then the options for this are numerous. However for titanium bolts or studs, we have less choice because titanium has a comparatively low modulus. We should note that a steel nut on a titanium stud or nut is likely to have a worse distribution of load than the case examined by Zhukovski.

Fig. 1 - High quality proprietary nuts, aimed specifically at the racing and high-performance market (courtesy ARP).

Written by Wayne Ward

There are, in the case of racing engines, a number of applications where the physical movement of a fastener from its design location can be disastrous. At this point we should note that if a fastener of any description, be it a nut, bolt or screw has moved from its preloaded position, the joint has failed already. This doesn’t necessarily mean the nut will fall off the bolt or the bolt fall out of a tapped hole. Any physical movement implies loss of preload and we know that correct preload is important in preventing fatigue failures of the fastener.

The ability of the joint and the system as a whole to cope with the loss of preload in any single fastener is a function of the amount of redundancy in the system. Safety-critical applications necessarily carry the penalties of a greater level or redundancy. These penalties are generally, in the case of bolted joints, higher component costs, higher assembly costs and greater weight. The racing engine in general has a much lower level of redundancy than many pieces of machinery because we value light weight, and so when we lose pre-load in a fastener in many race engine applications, we might reasonably expect serious damage to follow.

The obvious applications where, having a fastener rattling around unrestrained, are likely to cause a subsequent failure which is quite separate from any effect due to loss of preload in the joint. In fuel injected engines running an intake plenum, it is common to have quite a lot of hardware in the plenum, such as fuel rails and injectors, and commonly intake trumpets are held in place by small threaded fasteners - although I have seen two very different methods of fastening these which allows quick changes of intake length without the need for threaded fasteners. A loose fastener in the plenum has the possibility to be swallowed by the engine and ingestion of a screw can cause catastrophic valve and piston failure, a miserable end to a race day and a big bill.

Another application where loose screws are dangerous is in the crankcase. In many race series, people look to minimise the frictional losses in the crankcase due to oil-shear by trying to keep, as far as possible, the oil and the rotating pieces apart. This often involves the fitting of various scrapers and screens etc. and these are often held in place by small fasteners. Whilst in many wet-sump applications, this might not be critical, dry sump engines may suffer serious damage.

Written by Wayne Ward.

Ultrasonic methods are very accurate but don't seem to be in widespread use in motor sport. There are three other methods which seem common; tightening to yield using operator feel, measuring torque and calculating the tension from a mathematical relationship, and 'torque plus angle' where the bolt is tightened to a certain torque, then tightened by an additional angle.

When we tighten a bolt or nut, there is a shear stress due to torsion in the bolt; this is known to relax to a large extent once the applied torque is removed. So, if we tighten to yield, the torsional stresses will relax, allowing the material to go back below yield (ref 1), having already work-hardened the material slightly (be careful that what you feel is NOT the threads being pulled out of the casting!). If the service loads are low, we can be sure that the bolt will not surpass yield in service.

There are two main methods of calculating the relationship between torque and tension. One is to calculate this from first principles using assumed or measured values of friction for the nut face (or bolt face) contacts and the thread contact areas, the thread pitch and other dimensional details. The method is felt to have an accuracy of ±15 - ±30%, with the lower figure being based on close dimensional control of all parts involved and accurate data on friction. Friction data can come from published sources, or from your own tests.

The second method is called the nut factor method. Use this method cautiously, unless you make measurements to calculate the nut factor for each joint. It is folly to accept a given nut factor value for 1/2" bolts in steel as an example, and expect it to give good results if the fasteners and lubricants used are not the same as used to generate the nut factor value. Changes in thread pitch and friction have an effect on the torque-tension relationship; be sure that you understand the limitations of the nut-factor method before using it.

Torque-plus-angle is a method where a 'snugging' torque is applied to take the joint beyond the initial non-linear portion of the torque-tension curve. Application of the tightening angle in stages also removes much of the incertitude due to relaxation. For example, 300 degrees of angle might be applied as 3 x 100 degrees.

We can mitigate the variation in pre-load by conditioning the joint before final tightening, especially when using new parts. Tightening the joint, and then releasing the load completely allows the parts to 'bed-in' before proceeding with the final tightening. Some fastener suppliers recommend this procedure and some engine builders do this on critical applications. Pre-load relaxation is a fact of life, even where best practice is followed. Published data (ref 2) considers that 5% loss of pre-load is an acceptable allowance, but 10% relaxation is possible.

References:
John H. Bickford, An Introduction to the Design and Behavior of Bolted Joints, Marcel-Dekker, 1981, ISBN 0-8247-1508-X
Space Shuttle - Criteria for Preloaded Bolts, NASA NSTS 08307, 1998

Written by Wayne Ward.

The forthcoming article in Race Engine Technology on the subject presents some simple examples of how variations in fastener stiffness can affect the levels of stress in service, and more importantly the mean stress and stress amplitude. The combination of mean stress and stress amplitude are important factors in determining fatigue life.

By decreasing the stiffness of the bolt in relation to the joint stiffness, we decrease the proportion of service load taken by the bolt. It follows from examining the equation for load coefficient that, should we also be able to increase joint stiffness, that we can achieve the same effect. A simple way to do this might be to make a longer joint by increasing the distance between the nut and bolt (note that this won’t have the desired effect if there is a strange load path, or if the joint is essentially two hollow cylinders of the same diameters as the nut face. This would clearly add mass to the assembly and might not be possible within the constraints that we have. Another method might be to increase the modulus of the joint materials. Again this might neither be possible or desirable. The change to a stiffer casting material may bring negligible benefits if the same kind of material is used and in many cases we will have to work with production castings where we have no control over the material. A change from cast aluminium to cast steel would be disastrous in terms of added mass and so that option is a non-starter.

A very simple way to achieve an increase in joint stiffness is to increase the volume of clamped material, and this can be done with a washer having a diameter greater than the bolt-head outside diameter or the nut-face outer diameter if a nut is used. Washers are often used with threaded fasteners, but a lot of people ignore their technical significance in decreasing stress. Many people think that a washer is simply to prevent damage to a casting or low-strength component by the action of tightening. The head of the bolt or nut can indeed gouge a casting owing to the relative movement, or become slightly embedded due to pre-load. A washer can indeed prevent this damage, but more importantly, if it is stiff enough and is larger than the head of the bolt, it can increase the joint stiffness by effectively clamping a greater volume of material. If we look at the diagram in the previous article on the subject of joint stiffness (Part 4) we can see a joint with the two cones of clamped material, where a thin washer has been used. Thin washers deform easily and we don’t see the benefit of increased joint stiffness. However, if we use a thick washer made from a stiff material, we increase the clamped volume of material and consequently the stiffness of the joint. In terms of the diagrams in the previously referenced article, we increase the diameter of the clamped cones. This can have a significant effect for little increase in mass.

Ignore washers at your peril, and beware of the pitfalls of using thin washers.

Written by Wayne Ward.

e = ?L/L

This month we shall look at the calculation of joint stiffness, and the good news is that the methods and the basic calculations are the same. The combination of stiffness of different members is dealt with in exactly the same way. Where last month the calculation was dealing with different stiffness sections of the same stud or bolt, owing to changes in geometry, the individual calculated stiffnesses this month are for the individual components making up the joint itself.

To recap:

1/kt = (1/k1) + (1/k2) + (1/k3) + …..

where:

kt is the total joint stiffness
k1, k2 etc are the stiffnesses of the individual joint members.

For instance in a joint where two plates are clamped between two washers by a bolt, we would calculate the stiffness of the individual clamped members (two plates and two washers) and then calculate the stiffness as a whole using the above calculation.

Of course, if the geometry of each part is complicated, we can further subdivide this and treat it in the same way before using this stiffness in the overall stiffness calculation.

In calculating the stiffness of a bar in tension we assume that the whole of the cross-section is under stress and over the whole of the length. This is pretty much true. But what about a deep plate or some other geometry where the clamped area does not cover the whole of the cross section perpendicular to the fastener axis? How much of it do we include?

The theory is that the load spreads out from underneath the head of the fastener (or washer) at a given angle, and that the clamped material is therefore a truncated right cone (or frustum) with a simple hole in the middle. This is where the maths becomes more complicated / boring, so we shall just go straight to the solution.

So, for a symmetrical joint, i.e. with plates of equal thickness clamped by washers of the same diameter, there is a truncated cone in the upper plate, and a mirror image of this in the lower part with their junction at the interface between the two plates. The diagram below shows the upper part of such a joint:

In joints which don’t display symmetry, the junction between the two cones will not be at the junction of the two plates, but this shouldn’t present a problem. The problem is still quite simple if the two plates are of the same elastic modulus. If they don’t have the same modulus, one of the cones will have to be split into two individual cones, each having its stiffness calculated using the above formula. Joints not displaying symmetry may have different thickness clamped members or perhaps it is a joint held together with a stud threaded into one of the clamped members, as per the diagram below:

Where a stud is used, joints are often calculated using half of the engaged thread length being used to define the lower surface of the cone, but this depends on the geometry of the joint, and to some extent the length and quality of thread engagement. Reference books point to the fact that, beyond the first few engaged threads, little or no load is carried by the other threads unless special measures are taken.

Written by Wayne Ward.

We examined the concept of load coefficients and looked at a very simple example of how the load coefficient affects the cyclic loads in the bolt and the joint.

For the first part of the article this month, we will look at how to calculate the component stiffness for simple geometries, to allow us to calculate the load coefficient with reasonable accuracy.

The simplest possible component for which we can calculate stiffness is a straight bar of constant cross-section and, luckily, this is pretty much what a lot of fasteners are. For more complicated fasteners, we can model these with reasonable accuracy as a number of constant cross-sections in series. If we want to imagine how this might work, we should think of our spring analogy with a stiff spring and a less stiff one placed in series (stacked one on top of the other). If we apply a load to them, they both experience the same load and both deflect, but the less stiff one deflects more than the stiffer one.

We might remember from physics or engineering classes the concept of elastic modulus (often referred to as Young’s modulus after the British engineer of the same name), which is basically a measure of stiffness of a given material. It is defined as the ratio of stress to strain.

E = s/e

where:
E = elastic modulus (units of Newtons per metres squared)
s = stress (units of Newtons per metres squared)
e = strain (dimensionless)

Stress is the ratio of force to normal cross-sectional area, and strain is the ratio of change in length to original length, i.e.:

s = F/A
e = ?L/L

If we substitute these definitions into the formula for elastic modulus, we get:

E = FL/A ?L

The component stiffness, k is the ratio of force to elongation or change in length, i.e.:

k = F/ ?L

Therefore, we can, by re-arranging the equations see that:

k = F/ ?L = EA/L

This makes sense as we can imagine that a bar of large cross-sectional area deflects less under a given load than a smaller one. Those who, through practical experience, understand how springs behave know that a longer spring of otherwise identical dimensions (same coil spacing, wire diameter and mean diameter) and material will deflect more under the same load.

For a more complicated geometry, for example where there are shoulders etc on the bolts (see the attached picture of a shoulder bolt), we simply treat this as three simple columns in series – the main part of the shank, the increased diameter and the thread, which behaves as a reduced diameter. Some people consider the thread diameter to be the minor (root) diameter, and others consider it to be the mean of the minor and pitch diameters. The second case is the more conservative, giving a larger load coefficient.

For two or more columns in series, with stiffness’ k1, k2 etc, the formula to calculate the overall stiffness is:

1/kt = (1/k1) + (1/k2) + (1/k3) + …..

Next month we will look at joint stiffness which will then allow us to calculate load coefficient.

Written by Wayne Ward.

, joint geometry or manufacturing methods. There is some very simple maths involved, and the units I use are SI, so apologies to our readers who use ‘English’ units. However, we all use the same version of physics, so there should be nothing to puzzle anybody.

We need to have a good understanding of the joint behaviour as it becomes loaded, in order to determine the loads in the fastener. First, let us imagine that the bolt has been tightened until the joint is just held together, i.e. there is no actual tension in the bolt, and no compression in the joint. If a force then acts to separate the joint, the whole of this force is taken by the bolt, whose behaviour in this situation is identical to that if it had been tensioned in isolation.

This situation changes dramatically when the joint is pre-loaded by tightening the bolt further before the joint is subjected to working loads. In this situation, we must consider the elastic behaviour of the bolt and the joint and treat them as simple elastic components, springs if you like. We might consider the bolt to have a stiffness k1, and the joint to have stiffness k2. If we imagine that these are concentric and that we impose a deflection on the two springs, we can see that the force developed depends on the familiar equation

F = kx

where:
F is the force in Newtons
k is the stiffness in Newtons per metre
x is the deflection in the joint

In this case:

F = k1x + k2x, or F = x(k1 + k2)

The part of the total force used which increases the load in the bolt F1 = k1x

We can therefore define a ‘load coefficient’ which will determines the cyclic loads seen by the fastener.

F = F1 / F, i.e. F = k1 / (k1 + k2)

where:
F is the load coefficient, which is dimensionless
F1 is the force in the bolt in Newtons
F is the total force in Newtons
k1 is the stiffness of the bolt in Newtons per metre
k2 is the stiffness of the joint components in Newtons per metre

From this simple equation, we can draw a very useful fact. We can say that the smaller the load coefficient, the smaller the proportion of cyclic load seen by the fastener. This can help to steer us in the right direction for materials choice, bolt geometry and joint geometry.

So, if we say that the bolt is tightened to a preload Fp, and that the joint is cyclically loaded in tension by a force Fc, we can determine the maximum and minimum loads in the fastener and, with further information, construct a Goodman diagram to satisfy ourselves that the parts should not fail by fatigue.

Fmin = Fp
Fmax = Fp + FFc

where
Fmin is the minimum force in the bolt in Newtons
Fmax is the minimum force in the bolt in Newtons

So, if we preload the bolt by 20 units of force, subject the joint to a further 20 units of force as a working load we assume and the bolt has a stiffness one-ninth of that of the joint, i.e. k2 = 9k1, we can calculate the load co-efficient, and loads in the fastener and joint:

F = k1 / (k1 + k2) = k1 / (k1 + 9k1) = 1/10 = 0.1

So, when the joint is tensioned by the 10 units of force, the bolt sees one-tenth of this, and the force at the joint face is relieved by nine-tenths of this.

For the bolt:

Fmin = 20
Fmax = 22

For the joint:
Fmax = 20
Fmin = 2

We can see that, despite the fact that we have added an extra 20 units of force to the joint (which was equal to the preload in the bolt), there are two conclusions which might surprise the newcomer to bolt calculations:

there remains some force clamping the joint together
the maximum force in the bolt is only increased slightly

These are important conclusions and we shall build on this in further articles on the subject of bolted joints and fasteners.

Written by Wayne Ward.

Developing Force:
If we initially imagine that the thread on the fastener is ‘unwound’, with the axis of the bolt vertical, we would find that the thread is basically a wedge wound around a shank. The action of winding the fastener into the joint is much the same as forcing a wedge into a slit in a piece of wood. It is easy to push the wedge in by hand initially and then, as the wedge contacts both surfaces, resistance is increased and so it is with the fastener. Initially, if the threads are in good condition, the bolt should wind in by hand (there are some special circumstances where this isn’t true, but we will ignore these at the moment) and once the joint becomes tight, resistance increases. Thereafter, for a simple, stiff joint, providing friction remains constant and there is no yielding, tightening torque and fastener load increase linearly. Providing that we know the relationship between torque and force, and we know the load required on the fastener, we can tighten it (either the nut or the bolt) to a given torque to produce this load.

The torque required to produce a given load is given by the relationship below:

T = 0.159P + 0.578µpdp + 0.5 µwdw

where:
T = torque (Nm)
P = thread pitch (m)
µp = coefficient of friction between the thread elements
dp = pitch diameter (m)
µw = coefficient of friction between nut face (or bolt underhead) and the washer face
dw = washer face mean contact diameter

Therefore we can see that the load depends on quite a number of factors. There are much simpler equations, widely used, which lump these factors together as a ‘nut factor’ in the relationship below:

T = kP

where k is the nut factor. I would advise that you avoid this relationship unless you always use exactly the fasteners to which the equation applies, for it ignores the effects of those factors which you are likely to want to change. In Formula One there are few standard fasteners used (maybe more in future!), and most are to a custom design. This is true in a lot of other racing formulae albeit to a lesser extent. Also, in Formula One, the relationship between torque and load is often determined experimentally or by analysis. However, for those without such luxuries, the longer of the two equations should serve you well.

Future articles will examine how to determine a realistic pre-load, the effects of cyclic loading, fastener design, manufacture and materials.

For those of you who would like to refer to a good text on design of bolted joints, I can recommend the books by John Bickford, of which the handbook is the most comprehensive. These books are sought after and you will find that second hand copies change hands for hundreds of pounds in some cases!

1. Introduction to the Design and Behavior of Bolted Joints
2. Handbook of Bolts and Bolted Joints

Written by Wayne Ward.

Receiving no support from the fastener manufacturer the engine manufacturer turned to independent mechanical design consultant Tom Sharp, for a solution.The first stage, as with any failure analysis, was to physically analyse the failed flywheel bolts which were bespoke hex head bolts made from a precipitation hardening stainless steel.The standard engine had only failed a few of its eight bolts and it was immediately apparent that in each case, failure had occurred on the third thread away from the bolt head, which in this case was the first active thread. However, upon examining a non-failed bolt from the same engine it became apparent that the threads had stretched; this was visually obvious with a mechanical pitch gauge.Subsequent investigation confirmed that the manufacturer had been following poor advice and re-using the flywheel bolts; this test had been their fourth application. Under certain circumstances that would be acceptable but test calculations proved that the bolts were yielding in the threads upon tightening.The solution for the standard engine was simply to use new bolts in the short term and to re-use them only once, but in future, re-designed bolts will be made which will stretch in the shank rather than the threads, and which will be tightened to around 75% of yield strength.The increased torque engine had suffered an entirely different failure, as this was the engine’s first ever use and all eight bolts had failed catastrophically. Close examination showed evidence of fatigue failure (the engine had been run in) and failure was attributed to chronic overload. Whether this was due to the increase in torque or torsional vibration loading remains uncertain, as time pressure prevented further investigation on this front.The designer’s proposal was to modify the crankshaft to increase the thread size of the bolts, and to produce a new bolt from multi-phase alloy; a Nickel-Cobalt-Chromium-Molybdenum alloy which has a yield strength of 2000 MPa against 1200 MPa of a precipitation hardening stainless.The replacement bolt was designed with a location diameter under the head and a reduced shank such that the majority of the bolt’s stretching took place in the shank. They also ensured stress in the threads was at or below the level of stress in the shank. After production, tightening trials were carried out to ensure that the bolt would consistently achieve a given pre-load. A settling torque followed by an angle technique was used and the tightening trials repeated the results achieved by the designer/manufacturer to within a few percentage points.The speed of the project was such that within three weeks of TSMD’s Tom Sharp receiving the failed bolts, the new bolts were fitted and running successfully.