machine design, Vol.4(2012) No.

1, ISSN 1821-1259
AXIOMATIC DESIGN OF SIGNAL ANALYSES
Milo MILOVANEVI1, * - Jelena STEFANOVI MARINOVI2 - Boban ANELKOVI3
1,2,3

pp. 53-58
Research paper

University of Ni, Faculty of Mechanical engineering, Ni, Serbia

Received (08.01.2012); Revised (25.01.2012); Accepted (05.02.2012) Abstract: Axiomatic design (Suh 1990, 2001) shows that the engineering of good designs can be taught as a science. There is a simple set of underlying rules, or design axioms, on which good designs are based. The design axioms provide the opportunity for teaching design as a science, in that it is based on some simple underlying principles. Taking advantage of this opportunity a systematic approach to teaching how to engineer and critically evaluate good designs has been developed. Three steps are used in introducing axiomatic design. First is providing motivation, which is attempted by observing that design is fundamental to all engineering. The purpose of traditional engineering analysis, which is emphasized in engineering studies, is to support design. Second is developing the concept that there are two simple axioms, independence and information, that govern design, just as Newton's laws govern mechanics. Third is observing that in order to apply the axioms designs must be decomposed into a hierarchical structure. This leads to stating that there are three essential elements to engineering design: the axioms, the structure, and the process for creating that structure. Key words: axiomatic design, signal analyses

1. INTRODUCTION
The objective of this paper is to present a simple, straightforward approach for teaching engineers the theory and applications of axiomatic design in less than a full course. Axiomatic design is especially worthwhile because the design axioms elevate engineering design to a science, governed by a few basic rules, from what has been a kind of an art integrated with engineering analysis. The importance of the effectiveness of teaching essential methods in engineering design has a broad and significant impact on adding value to the practice of engineering. In 1997, Campbell and Colbeck published an insightful critical literature review on teaching and assessing engineering design. They note that there are no articles about how to assess students' designs and that the literature on teaching engineering design is relatively new, begun about 1987. Campbell and Colbeck point out that the approach to teaching engineering involves practice solving open-ended problems, creativity, teamwork, teaching by coaching, and experience for offering evaluation of student work. They stress the need for more education on teamwork. Evaluation of engineering designs in the absence of an underlying science is problematic. Love in 1980 showed how design evaluation could be based on the degree of achieving objectives, or functions. Pugh in 1991, taking a broad view, which incorporates marketing and selling, as well as traditional engineering design functions, also bases evaluation on a student's level of success in achieving design objectives [17, 18, 19, 20, 21, 22, 23]. The University of Maryland has made notable contributions to the literature on teaching engineering design including Dieter's book (1991) which emphasizes an algorithmic approach to design, and discusses

evolution, iteration, trial-and-error, art and intuition in optimization of designs. Dieter also shows a method for evaluating designs based on the degree of achieving the functions. Parcover and McCuen (1995) cite the need for curricular improvements. They focused on the constructive learning method, the content of which is process oriented, focusing on individual problem solving. They emphasize experience, and iteration with prototypes and testing for evaluating designs. Zhang (1999) discussed a product driven approach to teaching design to hundreds of first year students, at the University of Maryland which emphasizes experience, one-on-one instruction, and is resource intensive. Decomposition can be a useful approach for a variety of engineering problems. Frindley et al. (1992) discuss the use of a kind of lateral decomposition, or product dissection, divided into three segments: function, performance, and manufacture. The purpose was to facilitate the use of benchmarking in the design process, rather than assisting the design process itself. In that work, it was enabling the application of benchmarking that provided the motivation for product dissection or decomposition. Their paper focuses on development and applications of the benchmarking process for classroom use [10, 11, 12, 13, 14, 15, 16]. In a landmark book for advancing a scientific approach to design, Suh (1990) put forth a proposition that good engineering design is governed by two axioms: axiom one -maximizing independence (or minimizing coupling), and axiom two - minimizing information (or maximizing the probability of success). This approach is refined and further illustrated with expanded applications of axiomatic design in a second book (Suh 2001). In Suh's axiomatic approach, designs are evaluated according to their compliance with the axioms, which inherently

*Correspondence Authors Address: University of Ni, Faculty of Mechanical engineering, Ni, Serbia, A. Medvedeva 14, 18000 Ni Serbia, milovancevic@masfak.ni.ac.rs

Milo Milovanevi, Jelena Stefanovi-Marinovi, Boban Anelkovi: Axiomatic design in signal analyses; Machine Design, Vol.4(2012) No.1, ISSN 1821-1259; pp. 53-58

incorporates the degree of achieving the functional requirements. The design structure of lateral decomposition into design domains and vertical decomposition into hierarchies defines the necessary parts of applying the axioms. Suh also defines the process of creating the hierarchy, by zigzagging between the domains at progressively more detailed levels of the hierarchy, as well as by a process of physical integration of the components. These are excellent books (Suh 1990, 2001), although they may be challenging for any teaching situation other than a graduate or advanced undergraduate course dedicated to axiomatic design. For teaching axiomatic design, Bathurst (2004), in a paper given at the third international conference on axiomatic design and intended for those already familiar with axiomatic design, emphasizes the importance of examples and adapting existing design practices to independence analysis. He cites the strength of axiomatic design in promoting a hieratical representation that facilitates communication, documentation, and evaluation. He also notes that high-level restructuring of the design process in industry is required to implement axiomatic design. An axiomatic approach to design allows engineering design to be taught as a science. As in other sciences, basic underlying principles with broad applications can be taught... These principles are applied to solving design problems and are used for evaluation. This approach has clear and important advantages over the traditional approaches, which rely on compliance with sometimes abstract algorithms to attempt to ensure good results. These approaches often include intuition, innate creativity, and evaluation by experience, teaching by oneon-one coaching, and a process based on time-intensive iterations of trial-and-error work. A good design method should be self-consistent and could therefore be used to design the content as well as the best approach for teaching engineering design.

3. MATHEMATICS
AxD is based on mathematics. Mathematical definitions and concepts are combined and adapted to en efficient product developement management tool. The most important mathematical building blocks are domians, functions, mapping, matrices and statistical density functions.

3.1. The design process

An ongoing design process can be described as a mapping of what we want to achive (design goal) on an domain that describes how this will be achived (design result). This model is to crude to be used as a design management tool. Axiomatic design use standardized domains to create structure and facilitate systematic decomposition of customer requirements. Axiomatic design is partitioning the design process as follows:  Design goal; (Satisfy customer needs)  Customer domain; {Customer requirements}  Functional domain; {Functional requirements}  Physical domain; {Design parameters};  Process domain; {Process variables} ;  Design result; (Products that satisfies customer requirement in a robust way) Rows in black are axiomatic design standard domains. Text in {} are domain elements. Each domain pair represent a mapping i.e. the upper represent a design goal (what) and the lower a design result (how).

3.2. Mapping and zigzagging in hierarchies

Designs with just one customer requirement are mapped with simple mathematical functions. However, most products have more than one customer requirement leading to mappings in martrix equation forms. These mapping equations are called design equations. Complex products and processes are usually of a hierachichal design e.g. sturctured in a system, machine, component hierarchy. You usually zigzag between domain pairs when you work your way down through this type of hierarchy. This facilitates a creative design process without design solution preferences. Example: Mapping the functional domain onto the physical domain we chose an overall physical solution to satisfy an overall function and subsequently decompose the physical solution to functions at next hierarchical level and then mapping these onto the physical domain to create a more detailed solution. This is repeated until we have an implementable design [5, 6, 7, 8, 9]. The above described structure can be considered the design process fundamental structure. It provides good traceability and, if the independence axiom (se below) is satisfied, good scalability characteristics.

2. PRODUCT DESIGN
Axiomatic design is a tool for designing products and processes exceeding customer expectations. The axiomatic design process is axiom driven assuring technology neutral and innovative solutions for increased competitive power. The customer requirements are decomposed in trees of independent and robust product functions successively translated to product specifications. Axiomatic design is widely applicable including design of simple products, design of integrated hardware, software, man /machine processes and design of organizations. Design of products and processes are usually performed as an iterative organic growing process based on knowledge en experience within and outside the design project organization. Design goals are usually set based on previous designs and customer requirements as defined by the designer. Axiomatic design (AxD) is a widely applicable method used to achieve a well structured and efficient design process. AxD is knowledge based and focused on a systematical identification and breakdown of customer needs to product specifications resulting in products with to the customer robust attributes. 54

4. AXIOM
Based on the domain structure above the design axioms are implemented to create robust high quality products with customer appeal. In axiomatic design two design axioms are postulated the independence axiom and the information axiom.

Milo Milovanevi, Jelena Stefanovi-Marinovi, Boban Anelkovi: Axiomatic design in signal analyses; Machine Design, Vol.4(2012) No.1, ISSN 1821-1259; pp. 53-58

Based on these theorems and corollaries can be derived. This presentation will only adress the axioms. If we mapp the functional domain onto the physical domain the independence axiom can be defined as follows: 1. The Independence Axiom: Maintain the independence of the functional requirements 2. The design equation can be written in matrix form as FR = A DP where FR is a vector with functional requirements, A is the design matrix with design coefficients and DP is a vector with design parameters. The coefficient matrix also called design matrix, should be either diagonal or triangular to satisfy the independence axiom. For all other design matrixes, the design is dependent witch, given there is a solution, require an iterative method to find a useful design solution. This is also true for other domain pair combinations. The product should satisfy the independence axiom in all parts including the chain from customer requirements via product design to process design and down through the hierarchical trees between the domain pairs. This usually tests the designers creative skills. The result is often unique and patentable solutions. Designs that satisfy the independence axiom are more robust, simplify the design work, result in a less complex design organization, simplify operation and maintenance and are easier to use as platforms when developing new products on previous design generations. 3. The Information Axiom: Minimize the information content of the design The information content for a functional requirement is calculated with the functional requirement tolerance limits and the corresponding system performance density function as inputs. When the whole denisity function is within the tolerance limits the information function has its minimum value 0. For designs with more than one functional requirement, designs with independent functional requirements have the lowest information content [1]. The information content is used as a design quality measurement. It is used to compare different design concepts and solutions to select the design that satisfy the customer requirements in the best and most robust way .

4.2. Industrial applications

Well defined and executed axiomatic design projects usually yield good results. Some examples: MuCell a new plastic material with lower weight while improving toughness and dimensional stability resulting in an entirely new industry; Patented connectors high reliability and low weight Tribotek inc started based on the patent; Asian vehicle manufacturer used axiomatic design for Lean and Value add in the production process resulting in drastic reductions in work in progress, cycle times and volume of transport.

5. BASIC ELEMENTS OF AXIOMATIC DESIGN

There are three basic elements that need to be understood to get started teaching axiomatic design; they are: the design axioms, structures, and processes. The axioms are decomposed into the rules or grammar that governs design, i.e., the two axioms, already mentioned above and the corollaries and theorems (Suh 1990, 2001). The structures are decomposed laterally (Fig. 1) into the design domains and vertically into the design hierarchies (Suh 1990, 2001). The processes consist of decomposition, by zigzagging between the design domains (Fig. 2) down to the "leaf levels" or "nittygritty," i.e., the most detailed or fundamental level of the design hierarchy, and physical integration or composition of the final design (Suh 1990, 2001). Functional domain What it dose? Functional requirements FR1 FR2 FRn Physical domain What it looks like? Design parameters DP1 DP2 DPn Process domain How it is made? Process variables PV1 PV2 PVn

Fig. 1 Lateral decomposition ot trie design structure into design domains, ine arrows represent mapping between the domains. Understanding the nature of the mapping is the objective of engineering analysis and experimentation (Suh 1990, 2001). The lateral decomposition (Fig. 1) has been called the playing field for design, and the axioms, corollaries, and theorems called the rules. While the domains are usually presented with the customer domain (not shown in Fig. 1) and the process domain, the design exercises concentrate on the relationships between the Functional Requirements (FRs) and the Design Parameters (DPs) in the functional and physical domains. The process domain is used as a check to see if a reasonable manufacturing process and process variables exist. This helps with concurrent or simultaneous engineering. Manufacturing process and tool design are recast as product design problems aligned with the appropriate customer FRs and DPs. 55

4.1. Axiomatic design and quality of design

Standardized Design For Six Sigma (DFSS) processes have been designed with axiomatic design as one of the key components. Axiomtic design contributes to the design quality through the design axioms but also through the clear and customer focused structure facilitating coordination of other quality enhancing tools. Among these tools are TRIZ, innovation and patent databases, design for x, statistically designed experiments and associated robust design tools, cost / benefit analysis, risk management, FMEA, traceability systems and cost analysis.

Milo Milovanevi, Jelena Stefanovi-Marinovi, Boban Anelkovi: Axiomatic design in signal analyses; Machine Design, Vol.4(2012) No.1, ISSN 1821-1259; pp. 53-58

Defining the FRs properly is essential for a good design. The final design cannot be better than the FRs. Engineers have a tendency to jump immediately to physical embodiments, DPs, in addressing design problems. The result, frequently, is that the FRs are not well defined. Understanding the difference between FRs and Constraints (Cs) is also essential.

5.1. Decompositions
At each level of the hierarchy the FRs must be collectively exhaustive and mutually exclusive (CEME). This means that the required functions are covered without overlap between the FRs. The overlap would result in coupling and violate the independence axiom. Also, the decompositions must logically flow from one level to the next. If an FR at one level, say FR1, is to guard against mechanical impacts, then it might be decomposed in one of several logical ways. One way would be to decompose it into guarding against mechanical impacts from different directions, so that, for example, FR1.1 guards against vertical impacts, FR1.2 is to guard against frontal impacts, and FR1.3 is to guard against lateral impacts. Another decomposition scheme would be to guard against impacts of different duration. This could substitute for the above decomposition or be useful in the next level of decomposition. In the case of guarding against shocks from different directions, while three lower level FRs for guarding against impacts from three different orthogonal directions appears to be decoupled, i.e., mutually exclusive, it is not clear if oblique impacts would be included; therefore, the exercise would require some checking to be sure that these lower level FRs are collectively exhaustive. In the initial decompositions, a common mistake is to decompose an FR into only one other FR. This results in either an incomplete decomposition, i.e., the lower level FR is not collectively exhaustive, or else it makes the lower level FR redundant, just a redefinition of the higher level FR. In the FR decomposition structure each parent must have at least two children, unlike the biological metaphor.

microcontrollers for more then a decade, and microcontroller PIC16F877A has been developed as symbioses of microprocessor (CPU), memory and periphery, there for PIC is acronym for (Peripheral Interface Controller) [3]. This microcontroller is based on CMOS technology with RISC architecture and implemented FLASH and EEPROM memories. Thus, PIC16F877A represents best compromise between price and technology. Main signal characteristics that have been chosen as main requirements for design of optimal micro system are: resolution, stability and repeatability of signal. In order to fulfill requirement of resolution an A/D converter is added, since microcontroller has 10-bit resolution, a MCP3204 12-bit A/D converter is connected via SPI connection protocol [4].

6. OPTIMAL MICRO CONFIGURATION

Optima micro configuration for vibration monitoring is designed based on axiomatic design regarding electronic components selection. Basic idea in optimal micro configuration design is to fulfill frame conditions that are required for pump aggregates vibration monitoring. In order to define all components of micro configuration it is necessary to define functional requirements of system and conditions in which system will be tested.

6.1. Functional requirement of micro system for vibration monitoring

Considering micro systems for vibration monitoring the signal characteristics determine system accuracy and vibration monitoring quality in general. Selection of the microcontroller that is base for micro system has been done primarily considering economical aspects of newly developed system that is PC (personal computer) dependant [2]. Microchip, company that is producing 56

Fig.2. Diagrams of signal repeatability testing by exponential signal

6.2. Signal repeatability analyses

Analyses of main, selected signal characteristics have significant influents in optimal micro system selection. Significant signal resolution has been obtained by

Milo Milovanevi, Jelena Stefanovi-Marinovi, Boban Anelkovi: Axiomatic design in signal analyses; Machine Design, Vol.4(2012) No.1, ISSN 1821-1259; pp. 53-58

introducing MCP3204 12-bit A/D converter and stability is insured by selecting leading electronic components manufactures and implementing their components. Thus, testing signal repeatability in laboratory conditions, on signal generators Tektronix 3102 and Tektronix dpo 4034, is method for selecting optimal micro configuration. Following diagrams represent part of intensive signal repeatability testing of selected optimal micro configuration. Signal repeatability has been tested by two types of complex signals: exponential and triangle. Exponential signal in repeatability testing is presented in Figure 2 in following order, a) original signal screen shot of referent testing signal, b) fast furrier transformation (FFT) diagram of collected signal from tested configuration, c) collected signal from tested configuration without transformation.

(FFT) diagram of collected signal from tested configuration, c) collected signal from tested configuration without transformation. This testing has proved that selected optimal micro configuration for vibration monitoring based on microcontroller has suitable performance regarding signal characteristics.

7. CONCLUSIONS
Using axiomatic design requires greater investment in the higher levels of the design than is the practice historically. Engineers are often uncomfortable with investing so much time in design work, without producing drawings. Managers need to realize the value of design effort at the higher levels. For example, definitions need to be formulated to facilitate exclusivity and decomposition. The goals of the design need to be clearly and succinctly stated. Often it is clear from the homework efforts that there is confusion between designing the product and creating the design process for designing the product. Both exercises can have value, but they should be distinct. Also there is sometimes confusion between designing the process and designing the tools to accomplish the process. These kinds of confusions may be only apparent when a clear, systematic, and scientifically rigorous design approach is used. In addition, axiomatic design provides clear and simple rules for evaluating the results of students' work, eliminating the need to get experienced designers to legitimate the evaluations. Created monitoring system has significant results in frequency vibration analyses regarding mechanical defects detection of the pumping aggregate

REFERENCES
[1] Milovanevi, M., Milenkovi D., Troha S.(2009). The optimization of the vibrodiagnostic method applied on turbo machines. TRANSACTIONS OF FAMENA XXXIII-3 (2009), Faculty of mechanical engineering and naval architecture, ISSN 1333-1124 pp. 63-71, Zagreb [2] Milovanevi, M., Miltenovi , Bani, M. (2008). Spectral analysis of the working order conditions for the engines on pumping power units, Monograph Machine design, Faculty of Technical Science, Novi Sad, ISBN 978-86-7892-105-6 [3] Milovanevi, Milo, Miltenovi, ore, Bani, Milan (2009). Microcontroller based method for rotary machines monitoring, Machine Design: On the occasion of 50th anniversary of the Faculty of Technical Sciences: 1960-2009 (ISSN 1821-1259); Faculty of Technical Sciences; Novi Sad [4] Milovanevi, M., Cvetkovi M. (2009). Primena nove generacije mikrokontrolera za analizu stanja radne ispravnosti pumpnih agregata. Nauno-struni asopis Istraivanja i projektovanja za privredu. ISSN 1451- 4117 UDC 33. Br.23/24. pp. 35-41. 57

Fig.3. Diagrams of signal repeatability testing by triangle signal Triangle signal in repeatability testing is presented in Figure 3 in following order, a) original signal screen shot of referent testing signal, b) fast furrier transformation

Milo Milovanevi, Jelena Stefanovi-Marinovi, Boban Anelkovi: Axiomatic design in signal analyses; Machine Design, Vol.4(2012) No.1, ISSN 1821-1259; pp. 53-58

[5] Milovanevi, M., Cvetkovi M. (2009). Aplikativni znaaj optimizacije vibrodijagnostikog modela kod turbo pumpi. Nauno-struni asopis Istraivanja i projektovanja za privredu. ISSN 1451- 4117 UDC 33. Br.25. pp. 41-48. [6] Milovanevi, M., Anelkovi B. (2009). Wind turbine condition monitoring and control, Proc. of 3nd Internat. Conference POWER TRANSMISSIONS 09. 25 26 October, 2009. pp. 85.-90. Chalkidiki, Greece [7] Nelson, D., McVaugh, J.M. (1976). The Dynamics of Rotor-Bearing Systems Using Finite Elements, Journal of Engineering for Industry, May 1976, pp. 593.- 600., Paper No. 75-WA/DE-19, [8] Nelson, H.D., Meacham, W.L. (1981). Transient Analysis of Rotor-Bearing System Using Component Mode Syntesis, The Gas Turbine Conference & Products Show, pp. 1-7 [9] Nordmann, R., Massmann, H. (1984). Identification of Dynamic Coefficients of Annular Turbulent Seals, Proceedings of Rotordynamic Instability Problems in High-Performance Turbomachinery, NASA Conference Publication 2338, College Station, Texas, pp. 295.-312. [10] Nordmann, R., Weber, D., Hilbrath, K., Meinel, A. (1990). StudienArbeit 8613/90/S24, Arbeitsgruppe Maschinendynamik, Kaiserslautern. [11] Nordmann, R., Weber, D. (1990). Dynamik Vertikaler Turbomaschinen mit Fluiddynamischen Koppelelementen, 1.Zwischenbericht ueber das Forschungvorhaben No. 136/8-1, Kaiserslautern [12] Osgodd, W.F. (1965). Mechanics, Dover Publications, New York.

[13] Pestel, E.C., Leckie, F.A. (1963). Matrix Methods in Elastomechanics, McGraw-Hill Book Company, New York [14] Rades, M.: Mixed Precession Modes of RotorBearing Systems, Schwingungen in rotierden Maschinen III, Viewig, 1995., pp. 153.-164. [15] Rades, M. (1997). Use of Monophase Modal Vectors in Rotordynamics, Schwingungen in rotierden Maschinen IV, Viewig, pp. 105.-112. [16] Sapunar, Z., Krpan, M. (1980). Dinamika, Sveuilina naklada Liber, Zagreb. [17] Schaffrath, G. (1969). Ein Vervahren zum Berechnen der vier Feder- und der vier Dampfungskoeffizienten von Radialgleitlagern, Forschung im Ingenieurwessen 35, Nr. 6, pp.184.-195. [18] Simon, U., Brommund, E.t, Periodische Bewegungen einer Pendelzentrifuge in einem mehreckigen Fanlager, Schwingungen in rotierden Maschinen IV, Viewig, pp.181-188. [19] Smith, I.N. (1994). Programming in Fortran 90, John Wiley, Chichester. [20] Smith, J.M., Griffiths, D.V. (1998). Programming the Finite Element Method, John Wiley & Sons, New York. [21] Someya, T. (1989). Journal Bearing Databook, Springer-Verlag Berlin, Heidelberg. [22] Souza, A.F., Garg, V.K. (1984). Advanced Dynamics, Modelling and Analysis, Prentice Hall International, New Yersey. [23] Suh, NP (2001). Axiomatic Design: Advances and Applications. Oxford University Press, New York, NY.

58