Disc Spring Design & Manufacturing Assessment

The DIN standard was developed from a paper written in 1936 and hasn't really been modified since then. There are a few mathematical short cuts applied. In those days when the material and size of disc springs was far thinner and smaller, this approximation was fine. But with greater material thickness, and greater diameters, this intrduces errors.

When designing a disc spring stack, some engineers will select the "best fit" from the list of "standards" available. Your assumption is understandably that these must surely meet the permissable stress values for σOM as defined by the DIN standard, which states that ths stress should not exceed the tensile strength Rm of the material used? So for SAE6150 / 50CrV4, this value is near to 1,600N/mm2. You can look it up yourself, you will be given the following explanation (at least one is given, because many other manufacturers have no clue whatsoever): "some of these springs exceed the permissible stress at the point OM. As these springs have been manufactured for years with this overall height l0, we have not changed the height. With these types of springs there is the possibility of slight setting in use."

  1. Make sure you recognise the circumstances in which you CANNOT rely on commerially mass produced disc springs.
  2. Make sure you recognise when a disc spring MUST have an additional SET removal and PRE-STRESS manufacturing step in order for it to operate as required.
  3. Make sure you appreciate that such disc springs MUST have an installation or minimum PRELOAD at around 20% of S0.
  4. When ordering such disc springs, question how long ago they were manufactured - these are not items for your lomg term stortage.
  5. Understand how your stack design has more constraints within it and these must be accommodated.

Do you have comfort in the design, dimensions and performance of the unit disc spring? Not all of the standard sizes that the dominating large manufactruers publish are necessarily suitable or in fact meet the documented definition of a disc spring.

  1. Group 1 or 2 Disc Springs? Are these in recognisable and familiar configurations and easily assigned to either of Series A, B or C?
  2. Can you personally quab=ntify why the Group 3 Disc Spring dimension is correct for the contact flats as well as the thickness reduced
  3. Has the consulting enngineer used one of the Popular Specials because that was what was availabl erather than because this is a the best engieeriing solution?
  4. Does this disc spring with contact flats meet the definition of a disc spring as per the standard?
  5. Do you see the contradicton embedded in the prior questions - if so you are in the vast minority.
  6. Are these disc springs to operate at/in ambient temperature suitable for 50CrV4?
  7. If you are using a material other than 50CrV5, do you know to what extent, stress, fatique can be calculated, what is relevant and useful within the DIN standard?
  8. Must you deliver the disc springs configured in stacks?

If the base disc spring is "öff the shelf" are you sure that it is ideal or just a "best fit" out of what is available in the market?

  1. Outer Dia to Thickness ratio - Defines series class (with h0 / t) and various tolerances, which differ by series classification.
  2. Outer to Inner Diameter ratio - This has no impact on the Load Curve Shape but drives permissible stress values ~ for a ratio of 1.5 this will be -2600 N/mm2), whilst for ratios of 2 and 2.5 these are -3400 and -3600 N/mm2
  3. Cone height to thickness ratio - This determines the shape of Load Curve - linear vs progressive. If this value is greater than 1.3 don't even think of using these in a stack.
  4. Stack Height to Disc Spring Outer Dia - Once this ratio is getting over 2, the stack will not perform predictably.
  5. Do you appreciate how this disc spring will perform in the ambient operating temperature?
  6. Are you sure that the base material used in your disc spring is correct for the application? For example, in electical envionments check that the magnetism doesn't lead to induction heating!
  7. Disc Springs can be made from other materials whose properties are important for a specific reason, but do not extrapolate fatigue and other data from the published standards!
  • The disc spring height, l0 and, the overall height L0of any stack using such a spring must remain unaltered
  • The spring load for a reduced thickness disc spring must be the same as for an unreduced spring at s = 0.75·h, where h is the free height of the unreduced spring.
  • The t (original thickness) : t'(reduced thickness) ratios must meet those defined in the standard. (Series A and B are 0.94, Series C is 0.96)
  • The DIN2092 design formaula provides an additional term K4 (set to unity where no contact flats are used, to help deal with contact flat calculations.
  • The required width of the annulus/contact flat is implicit in point c above.... it must be derived from the requirement that the forces of the two springs must be equal at 75%h0
  • If the width of the contact flat has been derived from the "rule" that the it equal to 1/150 of the Outer Diameter - OD, then this is WRONG
  • The thickness of the disc spring is easy to verify, for dimensional series A and B, t′ ≈ 0,94 × t, and in the case of dimensional series C, t′ ≈ 0,96 × t.
  • If the contact flat width is not correctly calculated and the reduced thickness is correctly reduced, then the disc spring is not valid

Stacks get designed around constraints that exist in the surrounding design, space being the obvious one. Many stacks are just copied from other designs and the flaws inherent in these are ignored. As stacks grow in size or configuration complexity, the performance drop off is non-linear

  • Again, check the stack height to disc spring 0 ratio - Below 1.5 results will be predictable, over this, ie. from 1.5 to 2.5 be careful and make sure that maintainance is good, over 2.5 and ypu should expect unpredictable results and breakdowns.
  • How many packets make up the stack? More than 6 complex packets and more than 15 simple packets is not advisable?
  • How many disc springs are configured in parallel in a packet? Disc springs in parallel provide an additionl source of friction WM, apart from that which occurs between the contact points where loadinf occurs WR
  • If a long stack cannot be avoided, then break the stack into sections, separated by suitably thick, machined to even tighter clearance with the guide elements, and make sure these are hardened to at least Rockwell C of 55 and polished.
  • Remember that we have conservation of energy, friction will manifest itself in frictional hysteresis - from the Greek "lagging behind", so when doing a cyclical laod and unload test, there will be a greater differnce at the test point when loading and unloading.
  • The frictional forces have the effect of increasing the spring load under loading and decreasing it when releasing.
  • Calculate the disc spring rate and if you can, keep it low.
  • Operationally, make sure that the stack is kept cleam, suitably lubricated, and that the guiding elements are properly hardened.

Many of the common or stand disc springs will experience3 set loss, because the induced stress during compression are greater than the tensile strength of the base material. Their manufacturen involves a tricky step where the set is removed. The manufactured disc spring height is higher than the required height. Afte heat teratment the disc springs are full loaded ntil they are flat, when released this original extra height is lost, and the correct height is acheived. But not without consequenes. Understand these anmd you will be OK, ignore them and you are asking for trouble.

  1. Over time, all disc springs will lose some of their energy capacity either through creep or relaxation.
  2. Creep is defined as the change in height - 0 of the disc spring, by some Δl. Thus the amount of free travel must reduce (l0 -s) and the mechanical values of the disc spring obvious must change.
  3. Relaxation is defined as the loss of spring load ΔF over time when the disc spring is compressed to a fixed position.
  4. The amount of relaxation permissable differs between Alloy and high Carbon Spring Steel so watch out for this.
  5. If the dimensions of your disc spring require the extra set removal step, make sure this has been done properly.
  6. Disc springs left in strorage or inventory will experience recovery and their height will increase over time.
  7. These disc springs must have a preload when they are installed in the stack, make sure this is sufficient ie 15-20% of the total travel available
  8. Preload means that you must check that the maximum operational design load does go over 75% of the travel.
  9. Set removal through scragging introduces residual stress into the disc spring, these are in general aligned in the opposite direction to stress induced by loading.
  10. There is no practical way to quantify these residual stresses, the SAE design guidelines for manufacture claim that pre-setting increases the permissable σI by a factor of 120%! so they are significant.
  11. The amount of set is a function of the amount that the calculated stress when flattened exceeds the indicated percent of tensile strength.
  12. The set removal correction factor is a funtion of maximum percentage of tensile strength before and after scragging is complete.
  13. Perform the scragging a few times, rather than relyon on a single compression
  14. Perform your tests at the calculated Test Length defined by the standard

Static and quasi-static loads are those that where the disc springs are subject to a constant laod, or there are occasional load changes with longer intervals, the total number of load changes can be up to 5000 cycles.

If your load is not static then it must be dynamic.

  • The standard demands that any disc spring when flattened completely, must then return to its free height l0 and by implication L0 for a stack, within the tolerances for Free Height defined for each group 1, 2 or 3.
  • Likewise, for the Force/Load of the disc spring the tolerances must meet what is defined by group 1,2 or 3 (with Group 3 having the most demanding - +/-5% at 75% of travel).
  • For static loads: The highest calculated stress at the upper inner edge of the single spring (cross-sectional point I) is the most critical. The stress at cross-sectional point I has the highest magnitude and thus determines the set loss of the spring.1
  • For dynamic loads, we are concerns with different stresses. Fatigue fracture always begins on the lower side of the spring. Fracture will begin at cross-section II or III depending upon which position has the higher cyclical stress level.
  • When assessing your design stress, go to the S = 75% of h0 data and check the higest of either the σII or σIII stress value (in N/mm2) - this is the largest calculated tensile stress on the lower edges of the spring. This is the most important value for designing loads subject to fatigue.
  • Goodman graphs for fatique are only useful for stacks of 10 max, and best for 6 only - symmetrical deflection along the stack. As the number of disc springs in a stack increases, the number of load cycles that can be achieved is reduced.
  • Goodman graphs assume a sinusoidal load variation, real life is quite different to this, and dynamic loads are more usually some form of Dirac Delta or impact/shock more typified by a Square wave function.
  • Published Goodman graphs in the standards and available literature are only for SAE6150 and SAE1075, so don't use these for INCONEL etc
  • Fatigue charts are generated from impirical testing where the envirnment is far better controlled for friction and the stability of the ambient temp etc.







Top