Minimum-Size Mixed-Level Orthogonal Fractional Factorial Designs Generation: ASAS-Based Algorithm

Abstract

Orthogonal fractional factorial designs (OFFDs) are frequently used in many fields of application, including medicine, engineering and agriculture. In this paper we present a software tool for the generation of minimum size OFFDs. The software, that has been implemented in SAS/IML, puts neither a restriction on the number of levels of each factor nor on the orthogonality constraints. The algorithm is based on the joint use of polynomial counting function and complex coding of levels and follows the approach presented in Fontana (2013).

Study on sliding layer of cross-tensioned concrete pavement

“…Researchers have found that the friction force is the summation of mechanical meshing and intermolecular forces from the surfaces of two contacted objects (Fontana & Sampo, 2013), as shown in Equation (8):…”

“…The main structure parameters affecting the COF of the SPS sliding layer are the sand fineness modulus (SFM), thickness of sliding layer (TSL), and thickness of polyethylene-plastic sheeting (TPS). An orthogonal factorial experiment design (Fontana & Sampo, 2013) was used to evaluate the influence of structure parameters on the SPS. Two structure types of sliding layers were considered in this study: one with a single layer of polyethylene-plastic sheeting over a sand layer (Type I) and the other with a double layer of polyethylene-plastic sheeting with a sand layer between them (Type II).…”

Tuning synthesis flags to optimize implementation goals: Performance and robustness of the LEON3 processor as a case study

“…If all factors are quantitative, then two-level designs with centre points should be employed [18]. Otherwise, existing algorithms, like the one proposed in [19], can be used to generate the required mixed-level fractional factorial design.…”

Dependability-driven Strategies to Improve the Design and Verification of Safety-Critical HDL-based Embedded Systems

“…If all factors are quantitative, then two-level designs with centre points should be employed [113]. Otherwise, existing algorithms, like the one proposed in [57], can be used to generate the required mixed-level fractional factorial design. In practice, though, orthogonal multilevel designs are not generated by off-the-shelf statistical suites (like Matlab), but are subject of custom implementation for a particular use-case.…”

“…Second, parameters in practice often can't be reasonably quantified at two levels. Despite the existence of proposals for generation of fractional factorial designs for multilevel parameters [57], these algorithms are non-standard and are not supported by offthe-shelf statistical suites. Furthermore, resulting designs still may be quite huge, thus exceeding the experimentation time budgets (especially in dependabilityaware context).…”