1 INTRODUCTION This document contains a summary of class notes made during lectures at the University of Pretoria. Please note that, because these notes were produced during lectures, that there may be typing and calculation errors in the document. We tried our best to limit this to the minimum. If you detect errors in the document, please notify the lecturer as soon as possible. 2 OUTLAY OF OUR CONTACT SESSIONS The figure on the next slide shows a layout of topics that will be dicussed during contact time with Michiel Heyns. Design of welds π Static design Variable amplitude loading ππππ Maximum principal stress, π1 or π3 Von-Mises: ππ£π = 1 π1 − π2 2 + π1 − π3 2 + π2 − π3 2 2 Tresca (Maximum shear stress): π = Maximum principal strain: π1 Buckling Creep π1 −π3 2 Fracture Mechanics Fatigue analysis ππ ππ Fatigue analysis Cycles Miner’s Δπ1 damage ππ π·= ππ rule: Fracture Mechanics Crack propagation ππ ππ Δπ2 π1 π2 Endurance, π, (cycles) ππ = πΆ ΔπΎ + ππ π Stress intensity: ΔπΎ = π½Δπ ππ[πππ. π] Critical crack size ο· Plastic collapse ππππ ≥ ππ¦ ο· Fracture πΎ = ππ½ ππ ≥ πΎπΌπ 3 Equations to calculate static stress Consider the geometry shown below. The I-beam of length πΏ is welded with weld of size π all around as shown. The dimensions of the geometry is indicated in the figure. What is the stress at Points A and B? πΉπ» π¦ π πΉπ£ A π‘π π‘π€ π» π₯ B 8 Solution: The stress equation on a cross-section is given as follows: ππ₯ π¦ ππ¦ π₯ πΉπ§ π= − + πΌπ₯π₯ πΌπ¦π¦ π΄ The average shear stress is: π π= π΄ In the example above, say: π» = 400 ππ, π· = 300 ππ, π‘π = 10 ππ, π‘π€ = 8 ππ. We need to calculate the second moments of AREA as follows: 2 π» 2 π − π‘π€ π» 1 π» π‘π 2 1 3 3 πΌπ₯π₯ = 2 × ππ ( ) + 4 × π ( − π‘π ) + 2 × (π» − 2π‘π ) π + 4 × (π‘π π ( − ) + π‘ π) 2 2 2 12 2 2 12 π πΌπ¦π¦ = 2 × 1 3 π − π‘π€ π π‘π€ 2 1 π − π‘π€ 3 π‘π€ 2 π π + 4 × (( π) ( − ) + (( ) π)) + 2 × (π» − 2π‘π )π ( ) + 4 12 2 2 2 12 2 2 × (π‘π π ( π 2 ) ) 2 H 0.4 W 0.3 t_f 0.01 t_w 0.008 d 0.008 Calculations: I_xx 0.000446 I_yy A 0.015872 I_zz m m m m m Mx My Fz Fx Fy m4 0.000151 m4 m2 0.000597 m4 500 Nm 0 Nm 0N N N A x_A y_A sigma= Shear 0.15 0.2 224 219 0 B m m Pa MPa 0.15 -0.2 -224 219 0 Average shear stress calculation on a round section according to the factored resistance formulae: ππ = 0.67ππ΄ππ’ = 0.67π 0.707ππΏ ππ’ = 0.67 × 0,67 0.707ππΏ ππ’ = 0.45 0.707ππΏ ππ’ πππ π = 0.67 Average stress on a round object subject to shear forces and torque: X π¦ π X X π π₯ ππΉπ¦ = πΉπ¦ π΄ ππ = ππ πΌπ§π§ ππΉπ₯ = πΉπ₯ π΄ X This is for average shear stress distribution. Normally we would solve the shear flow equation that will result in large shear force dependent shear stress on the neutral axis. However, it seems as if you were only introduced to average shear stress calculation. Do not hesitate to contact me if you need more information. 4 Variable loads in Excel The following table and graph was used to demonstrate calcuation of maximum, minimum, mean and standard deviation of a signal in Excel. Chart Title 0 100 200 -100 50 150 30 100 -20 80 50 150 0 -200 1 2 3 4 5 6 7 8 9 0 -50 Max 150 -100 Minimum -200 Mean 9 -150 St. Dev. 101.0445 -200 -250 10 5 Explaining Rainflow counting The rainflow counting process is, in a nutshell, as follows: 1. Calculate extrema: a. This is the turning points at the top and bottom sides of the signal. b. Also called Peak-Valley reduction c. In most cases the turning points are grouped together to simplify. i. This is done by specifying the stress intervals in which turning points must be calculated. 2. Rainflow counting: a. Output is: stress amplitude (πππ ), mean stress (πππ and No. of cycles (ππ 3. For weld fatigue: a. Calculate the stress ranges (Δππ = 2πππ ) and number of cycles, ππ From the count shown below, we have the following rainflow counted detail: From [MPa] 1 2 3 4 5 6 7 8 9 50 100 150 -50 30 -100 100 20 -50 50 0 1 3 -100 -150 4 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 Range Ranges No. of [MPa] [MPa] cycles 50 50 100 80 150 100 200 120 80 150 80 200 120 70 50 5 2 1 ni Chart Title 0 -50 0 50 -50 100 -50 30 -100 20 -50 To [MPa] 50 -50 100 -100 30 -50 20 -50 0 2 3 4 5 6 7 76 8 9 9 8 10 11 1 1 0.5 0.5 0.5 0.5 6 STRESS LIFE S-N CURVE The stress amplitude at 1000 cycles is estimated as π103 = 0.9ππ’π‘ for steel, for 50% probability of survival (50% probability of failure) in π. The stress amplitude at 106 cycles is: π106 = 0.5ππ’π‘ for the same probability of survival mentioned above. How do we model the S-N curve then? π103 = 0.9ππ’π‘ πΆ = 0.9ππ’π‘ π 103 π106 = 0.5ππ’π‘ πΆ = 0.5ππ’π‘ π 106 πβπππππππ: 0.9ππ’π‘ π 103 = 0.5ππ’π‘ π 106 0.9ππ’π‘ π ( ) = 103 0.5ππ’π‘ π103 π log =3 π106 3 π= π 3 log 106 π10 For fatigue calculations, we normally have the stress ranges with their associated number of cycles over the period which the stress was measured. So, we need to have a formula to give the material endurance, ππ , for every stress amplitude, ππ . π106 π 6 ( ) 10 πππ π103 ≥ ππ ≥ π106 π = { ππ ∞ πππ ππ < π106 Where you make provision for the stress concentration notch fatigue and other reductions factors, remember that: 0.9ππ’π‘ π103 = πΆπ ππ§π πΆππππ πΆππππ πΆπππ πΎπΉ′ 0.5ππ’π‘ π106 = πΆπ ππ§π πΆππππ πΆπ π’ππ πΆππππ πΆπππ πΎπΉ Affecting factors: 1. Mean stress (Affects πΊπππ and πΊπππ a. Use Goodman mean stress correction. b. Calculate equivalent completely reversed stress amplitude, ππ for signal with amplitude ππ and mean ππ on a material with ultimate tensile strength ππ’π‘ . ππ ππ + =1 ππ ππ’π‘ ππ ππ = π 1− π ππ’π‘ 2. Size (Affects πΊπππ and πΊπππ . π 3. Loading effect (Affects πΊπππ and πΊπππ : ππ,πππππππ = π 0.7 4. Surface finish (Affects πΊπππ : Figure 1: Surface finish factors 5. Surface treatment. a. Effect depends on residual stresses produced. b. If residual stresses can be calculated from the treatment done, add this residual stress to the mean stress and do Goodman mean stress correction. 6. Temperature (Affects πΊπππ and πΊπππ : πΈπ 6 , πΌπΌπ π΅π’ππππ‘ππ 520 π π,2×106 = π πΈππ 2×10 1.0 πππ π ≤ 450 β πΆπ‘ = −3 1 − 5.8 π − 450 , πππ 450 < π ≤ 550 β 7. Reliability (Affects πΊπππ and πΊπππ : Table 1: Modification factors for different probability of survival 8. Corrosion (Affects πΊπππ and removes the endurance limit) Figure 2: Surface finish and effect of corrosion The S-N curve the becomes: π103 πΆ = πππ π π106 103 106 Endurance π Figure 3: Typical S-N curve for steels 6.1 Class problem from the slides For the class problem we have: ππ = π106 = 132.4 πππ π103 = 0.9 × 450 = 405 πππ From this: 3 405 log 132.4 = 6.18 The number of cycles to failure at 200 and 300 MPa. Remember to first compare the stress amplitudes with the modified 0.5ππ’π‘ π106 = πΆπ ππ§π πΆππππ πΆπ π’ππ πΆππππ πΆπππ πΎπΉ To prevent unnecessary calculations. You will be caught out by this in the tests. π= S_ut K_F K_F' S_10^3 S_10^6 m 450 MPa 1.7 1 405 MPa 132.3529 MPa 6.17638 Stress Endurance amplitude [MPa] 200 78 090 300 6 383 endurance limit: 7 STRESS LIFE FATIGUE EXAMPLE PROBLEM You have a steel part that is subjected to the stress spectrum shown in the table below, which was measured over a period of 2 years. Stress amplitude [MPa] 200 100 200 100 Stress No. mean of cycles [MPa] ni 100 10000 -200 50000 0 100000 0 10000 The material is steel with ultimate tensile strength 700 MPa. The specimen contains a notch with theoretical stress concentration factor 2. Assume a notch radius of 0.01 inch (0.254 mm). The machined specimen will be used at a temperature of 300 °C. What is the fatigue life of the component for a 95% probability of survival (5% probability of crack initiation)? Answer Estimate points on the S-N curve The stress amplitude on the S-N curve at 1000 cycles is given as: 0.9ππ’π‘ π103 = πΆπ ππ§π πΆππππ πΆππππ πΆπππ πΎπΉ′ The stress amplitude on the S-N curve at 106 cycles is given as: 0.5ππ’π‘ π106 = πΆπ ππ§π πΆππππ πΆπ π’ππ πΆππππ πΆπππ πΎπΉ Mean stress correction The following equation will be used: ππ ππ = π 1− π ππ’π‘ Notch fatigue factors From the lecture notes: 300 1.8 π=( ) × 10−3 πππβ ππ’π‘ 300 =( ) × 10−3 100 = 0.0072 Then: 2−1 π 1+ π = 1.58 πΎπ = 1 + Assume π = 0.01 πππβ. At 1000 cycles: Determine the factor q from the graph below and calculate: πΎπΉ′ − 1 π = 0.2 = πΎπΉ − 1 πΎπΉ′ = π πΎπΉ − 1 + 1 = 0.2 1.58 − 1 + 1 = 1.12 Affecting factors Surface finish (Affects πΊπππ : From the curve it is estimated at πΆπ π’ππ = 0.75. Temperature (Affects πΊπππ and πΊπππ : AT 300 °C, πΆπ = 1. πΆπ‘ = 1.0 πππ π ≤ 450 β −3 1 − 5.8 π − 450 , πππ 450 < π ≤ 550 β Reliability (Affects πΊπππ and πΊπππ : For a probability of survival of 95%, πΆπππ = 0.868. Corrosion (Affects πΊπππ and removes the endurance limit) The S-N curve for endurance π= π106 π 6 ( ) 10 π = { ππ ∞ Sut KF' KF S1000 Se m= 700 MPa 1.12 1.58 488.25 MPa 144.2089 5.66 3 π 3 log 106 π10 πππ π103 ≥ ππ ≥ π106 Csize Csurf CTemp Crel Cload πππ ππ < π106 1 0.75 1 0.868 1 Stress Stress No. CompletelyEndurance Damage amplitude mean of cycles rev stress [MPa] [MPa] ni [MPa] Ni Di 200 100 10000 233 65 507 0.15 100 -200 50000 78 INF 0.00 200 0 100000 200 156 848 0.64 100 0 10000 100 INF 0.00 Total damage = 0.79 Period = 2 years Fatigue life = 2.5 years 8 WELD FATIGUE 8.1 Problem Statement A flat section of thickness 28.95 mm is joined to a flat section of thickness 20 mm using a manual shielded metal arc welding process to produce the double V-groove weld of the butt joint shown in Figure 4. Both sections are from 300W structural steel and a normal-match electrode was used. The 2 mm root face has a root opening of 1 mm and the sections chamfered to produce a groove angle πΌ = 60°. The weld preparation was done in compliance with AWS D1.1:2008 as summarised in Table 3. The thick section is ground in the direction of the principal stress to produce a taper of 1:5. The height of the weld convexity is measured at 8% of the weld width with smooth transition to the plate surface. The welding procedure specification required the use of weld run-on and run-off pieces that were removed afterwards. The edges are ground flush with the surface with grinding marks in the direction of the axial stress, which is also the direction of the principal stress in this case. Welding was done from both sides and ultrasonic testing done to confirm the absence of sub-surface defects in the weld. The stress spectrum of the joint over a period of 2 years is as summarised in Table 2. For your information, the following has been included: ο· EN 1993-1-9:2005 Table 8.3. The design requires a safe life assessment method with high consequence of failure. The operating temperature is 250 °C and the surfaces corrosion protected using International Paint’s HT-10 epoxy paint. No post weld heat treatment was done. No post weld improvement (peening, grinding, dressing, etc.) were done to the weld detail. Table 2: Stress spectrum on the joint over a period of two years Number of cycles ππππ ππππ [MPa] [MPa] 200 100 100 000 50 -75 50 000 40 0 1 000 000 Figure 4: Tapered joint Table 3: AWS D1.1 double V-groove weld for butt joints (2008:96) Please answer the following (Please use EN 1993-1-9:2005 as reference): 1. What is the partial factor for fatigue for this problem? [5% of marks] 2. What other factors need to be taken into account in this problem? [5% of marks] 3. What is the detail category for this joint? [5% of marks] 4. What value must you take into account for the thickness effect in this case? [5% of marks] 5. Where do you expect the crack to initiate first in this joint? [5% of marks] 6. If the stress spectrum refers to a block loading applied to the joint over a period of 8 years, what estimate of fatigue life would you make for the welded joint for a probability of failure of 5% that is, for a probability of survival of 95%? [70% of marks] 7. What improvement in detail category is possible in this case with hammer peening? [5% of marks] Table 4: EN 1993-1-9 Table 8.3 for transverse butt welds (2005:22) 8.2 Solution Question 1: Partial factor for fatigue For a design philosophy of safe life assessment method with high consequence of failure, the partial factor for fatigue is πΎππ = 1.35. Question 2: Other influencing factors to take into account Temperature The new characteristic strength for the temperature compensated application is: ΔππΆ ΔππΆ,π»π = × 0.9 πΎππ Size effect The factor for size is: 1 0.2 ππ = { 25 ( ) π‘ π‘ ≤ 25 ππ π‘ > 25 ππ Question 3: Detail category Based on the information supplied, the detail category for this joint is 90. Question 4: Thickness effect No thickness effect need to be included, because, the crack is expected to initiate in the thinner section, which is 20 mm in thickness. Question 5: Point of crack initiation The crack is expected to initiate at the weld toe in the thinner 20 mm plate. Question 6: Damage and life calculation S-N curve The modified characteristic strength of the weld detail is: ΔππΆ ΔππΆ,πππ = ππ πΎππ π π This is the fatigue strength at ππΆ = 2 × 106 . The constant amplitude fatigue limit is then: 1 ππΆ π1 Δππ· = ΔππΆ,πππ ( ) , π1 = 3 ππ· The cut-off limit is: 1 ππ· π2 ΔππΏ = Δππ· ( ) , π€βπππ π2 = 5 ππΏ The endurance, ππ , for any stress range, Δππ , is then: ΔππΆ,πππ π1 ( ) ππΆ πππ 1.5ππ¦ ≥ Δππ ≥ Δππ· , πππ π1 = 3 Δππ π2 ππ = Δππ· ( ) ππ· πππ Δππ· ≥ Δππ ≥ ΔππΏ , π2 = 5 Δππ ∞ πππ Δππ < ΔππΏ { Damage and life calculation Based on the information provided to us, the fatigue life is 13.8 years. 90 ΔσD 44.2 MPa 1.35 ΔσL 24.3 MPa Detail category Partial factor for fatigue Thickness modification Temperature modification ΔσC,mod 1 m1 3 0.9 m2 5 60.0 MPa Endurance: NC 2.00E+06 Endurance: ND 5.00E+06 Endurance: NL 1.00E+08 σmax [MPa] σmin [MPa] 200 50 40 100 -75 0 ΔσR [MPa] 100 125 40 ni NR Di 100 000 432 000 50 000 221 184 1 000 000 8 245 044 Total damage = Period = Fatigue life = 0.231481 0.226056 0.121285 0.578823 8 years 13.8 years Question 7: Peening effect Based on the information provided to us, peening will result in a fatigue life of 28.8 years for 5% probability of crack initiation. 112 ΔσD 55.0 MPa Partial factor for fatigue 1.35 ΔσL 30.2 MPa Thickness modification 1 m1 3 0.9 m2 5 Detail category Temperature modification ΔσC,mod 74.7 MPa Endurance: NC 2.00E+06 Endurance: ND 5.00E+06 Endurance: NL 1.00E+08 σmax [MPa] σmin [MPa] 200 50 40 100 -75 0 ΔσR [MPa] 100 125 40 ni NR 100 000 832 550 50 000 426 266 1 000 000 24 607 671 Total damage = Period = Fatigue life = Di 0.120113 0.117298 0.040638 0.278048 8 years 28.8 years