Mohammed Musharaf Hussain

SIEMENS Q-RelayRedesign Project

Siemens Mobility industry project · Q-relay, BR930

Siemens' Q-relay had three faults, and all three came from the phenolic laminate its operating arms and adjustment cards are cut from. Our brief was to find a replacement that fixed all three, cost under £0.50 a part and could be recycled. The answer was unfilled polyetherimide (PEI), but the material that looked best on paper was not the one we chose.

01The root causeSiemens, BR930

The Q-relay is a plug-in signalling relay on British railways, built to the BR930 standard. Its operating arms and adjustment cards are cut from Tufnol, a phenolic laminate, and three faults followed:

Moisture linked them. In humid air, the phenolic and its cellulose filler let a current pass between contacts sitting close together, and silver migrates across the gap until it forms a bridge. The bridge is a short circuit, and a wrong-side failure in a signalling relay sends the wrong signal, so this was a safety problem before it was a quality one. Blanking added a second fault: Tufnol's layers delaminate under the punch and shed loose fibres. Together these gave us one hard constraint: the replacement could contain no phenolic or cellulose filler.

02Screening the fieldrequirements, EduPack, Ashby chart

We set the requirements before looking at any material: part cost under £0.50 including processing; water absorption under 56 mg at 1.6 mm thickness (ASTM D570); operation from −40 °C to +60 °C with self-extinguishing behaviour (BR930); recyclable; RoHS compliant; good electrical insulation; and ten million operations without failure. Phenolic and cellulose-filled materials were excluded outright.

In Granta EduPack we screened for strength, stiffness, fatigue and temperature, then ranked the survivors by cost per unit mass divided by tensile strength. Three contenders remained.

Price against tensile strength for materials in EduPack. Materials below the diagonal give more strength for the money. The three contenders are highlighted.

03Three contenders, one obvious winnertest data, decision matrix

PET looked like the answer. Its tensile, flexural and torsion results were the highest of the three, and it passed every requirement. Its weakness was recyclability: glass-filled, flame-retarded PET cannot be recycled, and Siemens requires a recyclable part. Polyester SMC is a thermoset that cannot be remelted, so it is hard to recycle. PEI passed every requirement too, with lower strength than PET but enough margin to spare.

We weighted the criteria by how much they mattered to Siemens: BR930 compliance (6), electrical resistivity (5), mechanical strength (4), water absorption (3), cost (2), and RoHS and recyclability (1 each). PET totals 95, PEI 82 and SMC 66, so the matrix favours PET. We chose PEI anyway, because recyclability was a requirement, not a preference. That is the one call where the matrix and our recommendation disagree, and it is deliberate.

Weighted totals, higher is better. Recyclability carries a low weight in the matrix, so it is treated as a hard requirement rather than a score.

PEI also absorbs more water than PET, 20.5 mg against 6.8 mg at 3.125 mm. That is well within the 56 mg limit, and the larger concern for silver migration is cellulose, which unfilled PEI does not contain.

04Testing the choicetensile, flexural, torsion

We checked PEI against the three requirements by hand, then in simulation. The test specimen is an adjustment card (47 × 7 × 1.5 mm, measured from a QNN1 unit) treated as a simple beam. A part passes if it carries at least 300 N in tension, more than 20 N in three-point flexure, and more than 0.3 N·m of torque at a 150° twist.

P.S. drag the slider to load the card.

PEI passes all three with margin. Flexure is the tightest, at 1.6 times its requirement, so a redesign would need to watch that test first. Finite element analysis (FEA) of the real CAD model in ANSYS agrees with the hand calculations, with an average factor of safety above 1 in tension and flexure.

05Keeping costs under budgetblanking, CNC, sourcing

Cost was the second hurdle. Extruding PEI sheet in-house would cost about £0.17 a part and need new equipment, so we ruled it out. Buying sheet is cheaper: a 1.575 mm unfilled PEI sheet from Boedeker Plastics costs £8.13 (March 2024), about £0.058 per part. CNC machining adds about £0.0525, putting the total near £0.11, well under the £0.50 limit.

Blanking on existing equipment should be cheaper still, but EduPack has no blanking cost model, so we used CNC as the benchmark. For the process itself, the fix was in the material: unfilled PEI has no layers to delaminate. For blanking we recommend heating the sheet to about 150 °C. Research on warm blanking of magnesium alloy AZ31B (Fazily et al., 2019) found that this suppressed loose particles and microcracks at the cut edge, and we adopted the same approach for PEI.

06The recommendationsolution

We recommend switching the operating arms and adjustment cards to unfilled PEI, cut by heated blanking or CNC machining. Each fault is removed at its source: