Pump bracket re-design tool saves Ford $1.1 million
Dean Palmer takes a look at how structural optimisation and FEA software helped Ford save $1.1m on a plastic steering pump bracket re-design project
Dean Palmer takes a look at how structural optimisation and FEA software helped Ford save $1.1m on a plastic steering pump bracket re-design project
Product development engineers at Ford Motors' site in Essex have saved the company $1.1m by re-designing a plastic power assisted steering (PAS) pump bracket using structural optimisation software.
Speaking at a recent software conference at the National Heritage Museum, David Wieloch, CAE analyst at Ford, said that "owing to reasons beyond Ford's control, the piece price of the plastic bracket had increased over the years to the point that it was excessively high and needed to be addressed." Therefore, Wieloch and Ford's front-end accessory drive (FEAD) systems engineer Ismail Taslim, initiated a review of the pump bracket design.
The bracket assembly is an important one as it appears on Ford's Sigma engine, which is used in the company's current Fiesta, Mazda 2 and Focus models and assembled at plants in Saarlouis, Valencia and Neihl.
Following the design review, a request was made for finite element analysis (FEA) work to be carried out on producing a totally new, alternative pressure die cast aluminium design. Having recently completed training on Altair Engineering's 'Optistruct' structural optimisation software, Wieloch said a decision was made to apply structural optimisation to the investigation "to produce a design with optimum material layout".
He explained further: "The original plastic steering pump bracket was a single piece, thin-walled polyamide moulding around brass and steel inserts. The whole bracket was two stiff half-box sections, held together by a flimsy connecting omega-shaped strip of material. The bracket didn't have any overall rigidity until the power steering pump was bolted to it. For dimensional accuracy, the bracket was located on a special jig fixture and the pump was then bolted down onto it. This assembly was then offered up to the engine and bolted to it as a combined unit." The purpose of the bracket was to act as a spacer between the PAS pump and engine.
According to Wieloch, the plastic bracket piece price had risen over the years "following the collapse of the original manufacturing supplier". Ford's current supplier was looking to increase the price further and Wieloch explained that, "by today's standards, the bracket was bulky and excessive in weight when compared to similar aluminium brackets. Also, the bracket was not the industry standard for today's brackets, which is aluminium, making this a very difficult part to resource competitively."
After discussions with systems engineer Ismail Taslim, Wieloch several key issues arose. The bolt footprints of the PAS pump to bracket and the bracket to engine were fixed. This kept the pump in the identical position as before, preventing the need for tooling for a new FEAD belt and eliminating re-validation of the completed new FEAD.
But, as Wieloch explained, "The height of the bracket to engine bosses could be reduced in depth. The new bracket weight and packaging targets were not to exceed the current design and the minimum current NVH target of 240 Hz was to be equalled or exceeded. And finally, the cost saving had to be significant enough to generate a full return on investment in under 12 months."
Before optimisation work began, the original plastic bracket targets were determined. Taslim said that a detailed FEA model of the bracket was modelled using Altair's 'HyperMesh' tetrahedral meshing option, complete with the steel and brass inserts. "In order to maintain results accuracy, a part model of the engine block and complete model of the 3kg cast iron PAS pump was also attached by the use of beam elements and rigid links to simulate the bolts," he continued. "FEA analysis was then conducted to determine the natural predicted frequencies of the assembly and the peak displacement due to the PAS pump belt load."
Using the existing plastic bracket CAD geometry and HyperMesh, a basic starting mesh FEA model was constructed from hexahedral elements. The model volume was constructed with both designable and non-designable volumes. As Wieloch explained: "The FEA model of the PAS pump model was added, along with the peak belt loading. For all optimisation runs, the engine model was removed and the bracket constrained at the base of the engine locating bolts. This was to reduce analysis time and file output size, because OptiStruct was running locally on the workstation."
The downside to this, he added, was that the predicted frequencies were higher, without the stiffness of the engine block. This was compensated for in the set up by setting a higher minimum target for the first frequency. The overall aim, he explained, "was to minimise the amount of material used from the designable region, resulting in an optimum material distribution that would satisfy its constraints while achieving targets better than the original bracket design."
The optimisation run was set up to minimise volume and reduce the first natural frequency to 350 Hz. This was higher than the 240 Hz target but, when the engine block model was incorporated in the final design work, a fall in frequency was seen. The new aluminium bracket was a pressure die cast part and consideration had to be made for the sliding direction of the die tools to form the bracket pockets. "OptiStruct allowed the addition of drawing constraints in the topology set up. This prevented the generation of totally infeasible designs," said Wieloch.
Following the optimisation run, the results were loaded back into HyperMesh for review. The first predicted frequency of 350 Hz was achieved with the minimised volume.
After confirming the results to Taslim, Wieloch said the FEA mesh was then converted to useful CAD data by creating surface data on the optimised surface. "This was achieved using OSSmooth in HyperMesh. The generated surface data was then exported from HyperMesh as IGES format, usable by our CAD and CAE engineers.
"We used this as a starting guide to generate an enhanced model that included die cast constraints such as strengthening ribs, draws and pockets. The first predicted frequency was 338 Hz."
After reporting on the final enhanced design to Taslim, Wieloch said a key decision was then made to leave the bracket as a two-piece design, not linked by thin connecting strips or feeders for the die. "This would reduce potential warranty costs from damaged joining strips and cut down on material, resulting in a lighter bracket. Having two separate pieces also meant cheaper packaging costs and less storage space was required. Also, the assembly process of the new bracket is identical to the original bracket."
The enhanced design was manually surfaced and issued to several aluminium suppliers for quotations. Wieloch continued: "The final step was to model and analyse the chosen supplier's design and compare this with the results of the original bracket. The supplier's IGES models of the new brackets were modelled and the PAS pump and engine model attached."
According to Wieloch, the weight of the original plastic bracket was 396g, compared to 289g for the aluminium design, a saving of 107g. He concluded: "OptiStruct is a powerful tool for creating optimised designs. The mass of the bracket fell by 27% while the performance of the first mode improved by 29.6%. The peak centre line and pulley displacement also reduced by 47% and 46% respectively.
"The OptiStruct results are not the finished solution, but were used as a guide to generate potential real world designs. CAE analysts should have at least a reasonable experience of product design, manufacturing skills and CAD. However, the potential payback to any major manufacturing company is considerable. Final calculations put the first year's saving for the new bracket design at more than $1.1 million. The cost of the original bracket was three times higher than the new one."
Pointers
* Ford saved a total of $1.1m by re-designing a plastic bracket on a power assisted steering pump for its Sigma engine
* Structural optimisation software was used to minimise the amount of aluminium used in the new design