Under-bolted hydraulic cylinder end cover: retorque, derate, or redesign?
A hydraulic press builder is commissioning a 160 mm bore cylinder that clamps a die-casting mould. Maximum working pressure is 25 MPa, and clamp pressure cycles from 0 to maximum about 1,200 times per shift. The blind-end cover is retained by 6 off M16 class 10.9 bolts (stress area 157 mm², proof load 130 kN each) screwed into tapped cast-iron threads and tightened with a calibrated wrench to the drawing value of 190 N·m using MoS2 grease, K ≈ 0.17. On the first pressure test the cover weeps oil at 18 MPa. The commissioning engineer re-checks the bolts and finds them at 150–170 N·m. The drawing carries a note: 'retorque after first shift'. The bolts are used ones salvaged from a scrapped unit of the same type. The customer's production start is tomorrow, and the project manager asks whether the joint can simply be retorqued to 190 N·m and run at full pressure, be derated, or must be changed.
1. Calculate the pressure load, the preload actually produced by 190 N·m, and the resulting margin. At what pressure does the joint open, and how does that explain the leak at 18 MPa?
Read the worked answer
Working load. Piston area A = π/4 × 0.160² = 0.02011 m², so F = 25 × 10⁶ × 0.02011 = 503 kN (84 kN per bolt).
Preload from torque. F = T/(K·d) = 190/(0.17 × 0.016) = 69.9 kN per bolt, i.e. 70/130 = 54 % of proof load. Total clamp = 6 × 70 = 419 kN, only 0.83 × the working load.
Consequence. Ignoring stiffness effects, the joint separates at p ≈ 419 kN / 0.02011 m² ≈ 20.8 MPa, before the rated 25 MPa. With the measured 150–170 N·m the preload is 55–62 kN per bolt, giving 331–374 kN total - essentially equal to the 18 MPa load (362 kN). The seal therefore loses contact pressure exactly where the leak appeared. Torque scatter with grease (±25 %) plus embedment and used-bolt relaxation means some bolts are lower still, so the joint opens locally and weeps.
The joint is under-bolted, not merely under-torqued. A pressure-containing joint with a cyclic load normally needs preload of 1.3–1.5 × working load, i.e. 654–754 kN, or 109–126 kN per bolt (84–97 % of proof) - not achievable with 6 × M16 10.9. Increasing torque alone cannot fix this; it would only risk thread stripping in the cast iron.
2. The project manager wants the joint retorqued to 190 N·m and run at 25 MPa tomorrow. What is your decision, what interim pressure limit can you defend, and what permanent change do you specify?
Read the worked answer
Decision: do not run at 25 MPa. There is no calculation that supports 25 MPa with 6 × M16 10.9 at 190 N·m, and a leak on a press clamp cylinder is both a fluid-injection hazard and a production-stoppage risk.
Defensible interim limit. With new bolts, correct lubricant and a calibrated wrench, usable clamp = 419 kN; allowing a factor 1.3 against separation, permissible load = 419/1.3 = 322 kN, so p = 322 kN / 0.02011 m² = 16.0 MPa; set the relief valve at 15 MPa for margin and record it as a temporary restriction.
Mandatory conditions for the interim state: replace all six bolts with new 10.9 (never reuse yielded bolts - they have already lost preload and may have damaged threads); clean and lubricate threads consistently; verify preload by torque-plus-turn or ultrasonic/elongation check on at least two bolts rather than torque alone; retorque after the first shift and after the first thermal cycle; and monitor for weeping.
Permanent fix. 8 × M20 10.9 (As = 245 mm², proof ≈ 203 kN): 0.75 × 203 × 8 = 1,218 kN ≈ 2.4 × working load, torque ≈ 0.17 × 152 kN × 0.020 m ≈ 520 N·m. Confirm spot-face diameter, cast-iron thread engagement (≥ 1.5 d, more for grey iron), cover stiffness, and O-ring groove dimensions; then pressure-test at 1.5 × design and re-issue the drawing with the correct torque and no retorque note compensating for a soft joint.
3. The clamp cycles about 1,200 times per shift. Explain why a joint that separates is a bolt fatigue problem, even though 25 MPa is well below the proof load.
Read the worked answer
Closed joint. When the preload is high enough that the cover never separates, the bolt only sees a fraction of the external load: ΔF_bolt = C × F_applied, where C = k_bolt/(k_bolt + k_joint) is typically 0.2–0.3 for a bolted cover. Per bolt, ΔF ≈ 0.25 × 84 = 21 kN, so the alternating stress is of order 21,000/(2 × 157) ≈ 67 MPa - comfortably below the fatigue strength of a rolled 10.9 thread with a large preload keeping the mean stress high but the cycle small.
Separated joint. Once the cover lifts, the bolt must carry the full 84 kN per bolt on top of its 70 kN preload, peaking near 154 kN - above the 130 kN proof load. Every clamp cycle then yields the bolt slightly. Three things follow: (1) the bolt elongates plastically, so preload decays and the joint opens at even lower pressure; (2) the thread root at the first engaged thread sees a stress range of roughly 84,000/157 ≈ 530 MPa, far above the endurance limit, giving finite life of the order 10⁴–10⁵ cycles; (3) the cover hammers the O-ring, which extrudes and fails.
At 1,200 cycles per shift, three shifts and 250 days, that is about 9 × 10⁵ cycles per year - bolts would begin breaking within months, and the failure would look like a 'sudden' fatigue fracture rather than the slow leak seen today. This is why the correct answer is a joint redesign, not more torque.