A geometric AccurAcy design method of multi-Axis nc mAchine tool for improving mAchining AccurAcy reliAbility

The reliability of machining accuracy is of great significance to performance evaluation and optimization design of the machine tools. Different geometric errors have various influences on the machining accuracy of the machine tools. The main emphasis of this paper is to propose a generalized method to distribute geometric accuracy of component for improving machining accuracy reliability under certain design requirements. By applying MBS theory, a comprehensive volumetric model explaining how individual errors in the components of a machine affect its volumetric accuracy (the coupling relationship) was established. In order to reflect the ability to reach the required machining accuracy, the concept of machining accuracy reliability is proposed in this paper. Based on advanced first order and second moment (AFOSM) theory, reliability and sensitivity with single failure modes were obtained and the model of machining accuracy reliability and the model of machining accuracy sensitivity with multiple failure modes were developed. By taking machining accuracy reliability as a measure of the ability of machine tool and taking machining accuracy sensitivity as a reference of optimizing the basic parameters of machine tools to design a machine tool, an accuracy distribution method of machine tools for improving machining accuracy reliability with multiple failure modes was developed and a case study example for a five-axis NC machine tool was used to demonstrate the effectiveness of this method. It is identified that each improvement of the geometric errors leads to a decrease in the maximum values and mean values of possibility of failure, and the gaps among reliability sensitivity of geometric parameter errors improved also decreased. This study suggests that it is possible to obtain the relationships between geometric errors and specify the accuracy grades of main feeding components of mechanical assemblies for improving machining accuracy reliability.

sciENcE aNd tEchNology

Introduction
Machining accuracy is critical for the quality and performance of a mechanical product and is an important consideration for any manufacturer.It is influenced by machining errors belonging to several categories, e.g.geometric errors caused by mechanical-geometric imperfections, misalignments, wear of the linkages and elements of the machine tool structure, by the non-uniform thermal expansion of the machine structure, and static/dynamic load induced errors [8].Geometric errors include pitch errors of the lead screws, straightness errors of the guide ways, angular errors of machine slides, and orthogonal error among machine axis [3].Because its contribution to 30% of the total error so it is given special consideration through the configuration and allocation of appropriate dimensional errors in the design of machine tools with satisfactory machining accuracy [19].Machine tools are usually made by several assembling parts, and the dimensional and geometric variations of each part have to be specified by tolerances which guarantee a certain level of quality in terms of satisfying functional requirements [20].A as a result, the distribution of accuracy of machine tools is a problem of distribution of tolerances of these geometric errors.However, this practice of allocation has not been developed and applied in any systematic manner to the design of machine tools.
To enhance the machining accuracy of CNC machine tools, there are two steps included in the accuracy design.The first is accuracy prediction, which refers to forecasting volumetric errors of machine tools based upon the known accuracy for updated and maintained parts, and then predict the machining accuracy of work piece.The other is accuracy allocation, which is to obtain the accuracy of updated and maintained parts from the preset total accuracy of machine tools, and let the accuracy of parts reach optimal scheme [15].Before accuracy distribution, a error modeling is crucial to maximize the performance of machine tools, and robust and accurate volumetric error modeling is also the first step to correct and compensate these errors [5], therefore, a model explaining how individual error of the components of a machine affect its volumetric accuracy is crucial to the accuracy distribution approach and it is one aspect of importance of this paper.
Geometric errors for each of the assembled parts and components are random variables; the machining errors caused by such errors are also random [40].As a result, the dimensional and geometric variations for each error source are random and have to be specified by variance (or standard deviation) and the probabilistic nature of the errors produced on the dimensions of a manufactured part is taken into consideration in this paper.Uncertainty in parameters such as material, loading, geometry and the model exist in the process of structural design and optimization, and it results from data shortage, model simplification and human error [18].Some of these errors can be reduced by collecting more data, with a better understanding of the problem and by implementing strict quality control; however, others such as deviations due to random events cannot be reduced by the above means [6].To solve this problem, two main philosophies dealing with the uncertain deviations exist: 'Deterministic Structural Optimization' (DSO) uses safety-factors to accomplish the safety, while 'Reliability-Based Structural Optimization' (RBSO) takes the random character of the variables into consideration [27].Reliability-based design optimization simulates all the uncertain variables to the random variables or random process and minimizes an objective function under probabilistic constraints.The reliability of machine tool reflects the ability to perform its specified functions under the stated conditions for a given period of time and it is often studied by possibility of failure [26], so accuracy distribution based on the reliability theory can maintain and improve the level of quality in terms of satisfying functional requirements while keeping the maximum tolerances of these geometric errors.As a result, the development of a systematic method to realize accuracy distribution of machine tool based on reliability theory according to failure modes for improving machining accuracy reliability is the second very important aspect of this paper.
The rest of this study is organized as follows: In Section 2, the review of accuracy design of machine tool is given.Section 3 explains the process of modeling machining tool geometric errors.Section 4 presents the proposed method to realize accuracy distribution of machine tool.A case study is accomplished as an example in Section 5.The conclusions are presented in the last Section.

Accuracy prediction
Up to present, there are many researches on the error modeling technique to show the difference between the actual response of a machine to a command issued according to the accepted protocol of that machine's operation and the response to that command anticipated by that protocol [9].The development of modeling methods has been experienced many years and it turns out to be various kinds such as matrix translation method, error matrix method, rigid body kinematic, D-H method, model methods based on the multi-body system kinematics theory and so on.In 1973 Love and Scarr obtained the combined effects of the elemental errors in the machine tool and then developed the volumetric errors of a multi-axis machine by using the trigonometric technique [25].In 1977 a matrix translation method was reported and a calibration technique using three-dimensional metrology on a coordinate measuring machine (CMM) was presented by Hocken et al. [13].In 1981 Dufour and Groppetti reported the "error matrix" method to obtain error predictions by interpolation between the stored values [10].In 1982 Portman used rigid body kinematics to evolve an expression for the geometric error of a mechanism [30].In 1991, Kiriden developed a general model to understand the effects of component geometric errors on the kinematic chain of a machine and the volumetric errors in the work space [21].In 1993, they used the D-H convention to develop kinematic models for three types of machine.In 1995 a method based on direct consideration of the shape and joint transformations was put forward by Srivastava et al. [33].In 2007 Bohez et al. presented a new method to identify and compensate the systematic errors in a multi-axis machine tool [1].In recent years, multi-body system (MBS) theory is used to generalize and provide a unique systematic approach for its advantages such as stylization, normalization, versatility, and ease for computer modeling [36].There are many investigators have carried out error modeling research for complicated machinery system using MBS [41,23,44,39], mainly focus on designing and constructing a model to determine geometric error of machine tool and developing the key technique for compensation-identifying geometric error parameters.[43,17,4,7,22,2,11] introduce the methods of geometric error compensation, thermal error modeling, position error compensation, position-independent geometric errors modeling, volumetric error modeling and sensitivity analysis and establishing a product of exponential(POE) model for geometric error integration.

Accuracy allocation
In earlier years, many researches focused their attention to obtain the tolerance allocation on structural design.In 2005 Prabhaharan et al. introduced a kind of metaheuristic approach as an optimization tool for minimizing the critical dimension deviation and allocating the costbased optimal tolerances [29].In 2006 Huang and Shiau obtained the optimized tolerance allocation of a sliding vane rotary compressor's components for the required reliability with the minimum cost and quality loss [16].In 2007 Huang and Zhong established the sequential linear optimization models based on the process capabilities to release sciENcE aNd tEchNology the working tolerances, reduce manufacturing costs [14].Siva Kumar and Stalin [31] used Lagrange multiplier method to simultaneously allocate both design and manufacturing tolerances based on minimum total manufacturing cost.Isabel González et al. [12] developed a methodology to allow an automatic tolerance allocation capable of minimizing manufacturing costs based on statistical approach.Muthu et al. [28] used metaheuristic method to balance the manufacturing cost and quality loss to achieve near optimal design and process tolerances simultaneously for minimum combined manufacturing cost and quality loss over the life of the product.In 2010 K. Sivakumar, et al proposed a novel multi-objective optimization method to enhance the operations of the non-traditional algorithms and Multi-Objective Particle Swarm Optimization and systematically distribute the tolerances among various the components of mechanical assemblies [32].From the above literature, the previous researches on tolerance allocation mainly focus on structural design taking manufacturing cost or manufacturing process into consideration, besides, there are many of accuracy allocation in the field of hull construction, robotics, military application, and instruments [35,24,37,42,34].However, works on accuracy allocation of multi-axis machine tools are few.Reliability is a specification to measure the ability of machine tool to overcome a certain functions and reliability sensitivity reflects the influence of basic parameters to the possibility of failure, so reliability theory plays an important role in accuracy allocation of NC machine tools by determining the levels for these geometric parameter errors, what's more, multi-axis NC machine tools composed of various parts are complex structures and so they have multiple failure modes.In 1994 Dorndorf U proposed an error allocation approach to optimize allocation of manufacturing and assembly tolerances along with specifying the operating conditions to determine the optimal level for these errors so that the cost is minimized [9], it is regardless of reliability sensitivity and the model is a twoaxis machine.In 2013 Yu proposed a geometric error propagation model and reliability approximately model to by response surface method with error samples and improved the functions of machine tools by optimization of the sensitivity [40] with single failure model.As a result, the continuous effort lies on accuracy allocation of NC machine tools taking machining accuracy reliability and sensitivity with multiple failure modes into consideration.

Error modeling of NC machine tool
There are two important aspects in this study: the first is the development of a systematic approach to obtain geometric/kinematic errors on the kinematic chain of a machine tool.Another aspect, considered more important, is the proposing of the concept of machining accuracy reliability and the addressing a reliability and sensitivity analysis method in the multiple failure modes to realize the distribution of the standard deviation for the geometric errors.
A method called MBS is used to establish a machine tool geometric/kinematic error model showing the relation between the individual error of the components of a machine, and its volumetric accuracy is crucial to the allocation of standard deviation for the geometric parameter errors.In this paper, a 5-axis machine tool is used to analyze geometric errors and the geometric/kinematic error model is developed.This XKH1600 five-axis machining center is designed for leaf blade machining, configured as three linear axes X, Y, Z axes and two rotary axes A, B axis.The 3-dimension digital structure model of the machine tool is shown in Figure 1.
Taking the error factors and coupling relations of the various parts into consideration, based on the MBS theory, the five-axis machine tool can be abstracted into a multi-body system.The coordinate sys- sciENcE aNd tEchNology tem structure diagram is shown in Figure 2. The error model between K B and j B revealed, when the displacement and the displacement error are zero, k O and k Q coincide, k q refers to the initial position vector between j O ( j B 's origin) and k O ( K B 's origin), and ke q is the position error vector, including load error and thermal error.k S is the displacement vector between K B and j B , and ke S is the displacement error vector, including geometric error and dynamic error.When there is a displacement in one part or component of a machine tool, this displacement is the position increment.The lower body array is listed in Table 1.
The work-piece forming point in the work-piece coordinate system coordinate is: When the machine tool moves in ideal form, that means the machine tool is without error, the ideal forming function of tool forming point in work-piece coordinate system is: In this paper, the ideal forming function of tool forming point of this five-axis machine tool is: Machining accuracy is determined by relative displacement error between the tool forming point of machine and work-piece.During the actual machining process, the actual position of cutting tool point will inevitably deviate from the ideal location, which results in volumetric error.Table 2 lists the characteristic matrices of this 5-axis CNC machining center, including body ideal static, motion characteristic matrix ( ( ) ( ) ) and body static, kinematic error characteristic matrix ( ( ) ( ) ).As a result, the comprehensive volumetric error caused by the gap between actual point and ideal point in this paper can be written as: In and so on can be obtained from Table 2.

Accuracy allocation based on Reliability Theory
In Section 3, the development of a systematic approach to obtain geometric/kinematic errors on the kinematic chain of a machine tool has been finished.As a result, the concept of machining accuracy reliability should be proposed, besides, a reliability and sensitivity analysis method with multiple failure modes for improving machining accuracy reliability to realize the geometric errors allocation should be addressed.

The concept of Machining Accuracy Reliability and AFOSM Theory
The reliability of structure reflects the ability to perform its specified functions under the stated conditions for a given period of time and it is often studied by possibility of failure.In order to reflect the ability of machine tools to reach the required machining accuracy, machining accuracy reliability is proposed in this paper and can be defined as possibility of fulfilling the specified machining accuracy, expressed by the possibility of failure of machine accuracy.Besides, the probabilistic nature of the errors produced on the dimensions of a manufactured part is taken into consideration in this paper.In order to develop an approach to obtain the machining accuracy reliability and sensitivity of this five-axis machine tool, there is an assumption that the errors produced on a dimension are drawn from a Gaussian distribution.Our task of error allocation is therefore to determine the optimal levels for these geometric parameter errors according to the machining accuracy reliability.In order to overcome this task, Advanced First Order and Second Moment (AFOSM) theory is introduced at first.

S S P P S S S P S P P t T T T T T T T T T p
sciENcE aNd tEchNology Body static, kinematic error characteristic matrix ( ( ) ( )

3-4 Tool
sciENcE aNd tEchNology tivity coefficient, reflecting the influence of the random parameter i x acting on the σ z , so σ σ α , and then the reliability index and the possibility of failure are obtained as follows [26]: Compare with machining accuracy reliability as a specification to measure the ability of machine tool to overcome a certain functions, machining accuracy sensitivity reflects the influence of basic parameters to the possibility of failure, which can be used for improving and optimizing the basic parameters of machine tool.As a result, machining accuracy sensitivity plays an important role in determining the levels for these geometric parameter errors.Based on the AFOSM theory, the performance function

An AFOSM Reliability and Sensitivity Analysis Method with Multiple Failure Modes
Failure mode is critical to the reliability of any structure.The number of limit state equations divides failure mode into two parts: single failure mode and multiple failure modes.Single failure mode which means that there is only one limit state equation in the whole system or structure, in contract, multiple failure modes refers to multiple limit state equations in the system and its complexes lies in the logical relationship, correlation coefficient and joint probability density of a system or a structure with multiple failure modes [38].According to the logical relationship with the failure modes of a machine tool, this machine tool has multiple failure modes and so it is a serial system.As a result, this paper aims to the reliability and sensitivity analysis of such serial system.

The Narrow Bounds Method
The narrow bounds is the main method used for the reliability and sensitivity analysis of a serial system though it fails to obtain the certain valves of reliability and sensitivity, as a result, it is used for verification of the method introduced in this paper.
Supposing that there are "m" failure modes, the reliability P {F} and the reliability sensitivity ∂ ∂

P F
x i { } σ based on AFOSM Theory are expressed as follows [9]: In the above expressions, i F and j F are the events of failing with respect to the i Z and j Z condition respectively.
, β i and β j are the reliability indices with respect to the ith and jth failure conditions respectively, and ρ ij is the correlation between the two failure conditions, i q is positive when Z>0 and is negative when Z≤0. ( , ,..., ) In this way, and From the following equations, the reliability sensitivity can be obtained as follows: sciENcE aNd tEchNology

A Method for Reliability and Sensitivity Analysis with Multiple Failure Modes
For the failure to obtain the certain valves of the reliability and sensitivity by the narrow bounds, a method used for reliability and sensitivity analysis with multiple failure modes was introduced in this paper and the solution procedure roughly follows the approach developed by [38].
Given that there are "m" multiple failure modes in a system, the calculation of its possibility of failure is to obtain the integration of a multidimensional normal joint distribution.As a result, the possibility of failure of a serial structure with multiple failure modes is given by: As to a general serial system, the possibility of failure can be expressed as follows: In order to calculate the equation ( 12), the correlation coefficient is proposed to show the correlation with multiple failure modes.If there are two failure modes, the possibility of failure can be expressed as follows: 1 ) ( ) ... ( ) ( ) ...
In the equation ( 15), α Up to now, the possibility of failure with "m" multiple failure modes is obtained and the next work is to focus attention on the calculation of reliability sensitivity.For lack of space, a description of the complete theoretical analysis is avoided and only a conceptual description and important results are given. ...
... 12 (16) In the equation ( 16), ∂ ∂ ( ) σ can be obtain by equation( 8) and , In order to develop an approach to distribute this five-axis machine tool on machining accuracy, the method used for the machining accuracy reliability and sensitivity with multiple failure modes is introduced in this paper and the calculation of the reliability and reliability sensitivity is developed in MatLab software.Besides, the process of this accuracy distribution method in this paper can be shown in Figure 3.

Case Study
In the preceding section, a method of geometric accuracy design of multi-axis machine tool for improving machining accuracy reliability was introduced and a machining accuracy reliability model and a sensitivity model with multiple failure modes based on AFOSM theory were established.Previous to it, a method for relating the volumetric accuracy of a machine to the errors on individual links and joints in its kinematic chain was described and the error model can be expressed by the form: In this paper, the machining accuracy reliability limit state equations of the machine tool are subject to the expression as follows: From the equation (18), it is not difficult to notice that this machine tool has 26 failure modes in total, including 6 single failure modes, 12 double failure modes and 8 triple failure modes.
One of important aspect of the design requirements of such machine tool is that the probability of failure of position error less than 0.03mm should be no more than 5%.According to "Test code for machine tools-Part 1: Geometric accuracy of machines operating under no-load or finishing conditions" and "Test code for machine tools-Part 2: Determination of accuracy and repeatability of positioning numerically controlled axes", the values of geometric parameter errors of five-axis NC machine tool were set initially and were shown in Table 3.In order to improve the machining accuracy reliability of this fiveaxis NC machine tool, the models based on AFOSM theory addressed in section 4 was used.The process was divided into two steps: one was to determine some points by orthogonal sampling method in the X-, Y-plane of this machine tool to obtain the reliability, and the other was to realize accuracy allocation of all the geometric parameter errors of this machine tool.Five points 50, 225, 275, 325, 500 were chosen in X-axis and −225, −50, 0, 50, 225 were selected in Y-axis, as a result, there were 25 points in X-,Y-plane.The machining accuracy reliability model for accuracy allocation of machine tool addressed in this paper used to optimize errors was given by the expression (15), and its process of calculation was operated in the MatLab environment.According to the constraint conditions, if max ( ) 5% l were obtained though putting the vector of each point into the equation ( 5) and σ x i should be no more than the initial valves shown in Table 3.In order to verify the effectiveness of the method introduced in this paper, the possibility of failure of this machine tool based on the narrow bounds method was also calculated.The values of the possibility of failure of machining accuracy before improvement were acquired and shown in Table 4.
It is not difficult to notice that there are two important points in Table 4: one is that the method developed in this paper is verified because the values of possibility of failure calculated by such method are in the intervals obtained according to the narrow bounds method, and the other is that the initial values of the reliability of this machine tool failed to satisfy the design requirement (no more than 5% and 3% respectively).As a result, it is necessary to reallocate the initial values of geometric parameter errors in Table 3 according to the sensitivities of such errors shown in Figure 4.In order to satisfy the design requirement five times improvements have been developed.From the Figure 4 to Figure 8 we can learn that geometric parameter errors ∆β xz , ∆α yz , ∆α y , ∆γ x , ∆β x have the largest values of reli- ability sensitivity and have been improved as the changing objects in turn as a result, besides, the machine accuracy reliability (expressed by possibility of failure) with these five improvements have been cal-culated as shown in Table 5.It suggests that the whole process of accuracy distribution have been finished after the forth improvements.
Figure 9 illustrates the relation between the probability of failure and each improvement, it is clear that each improvement leads to a decrease in the maximum values and mean values of possibility of failure and such decrease becomes very tight after a certain improvement.Figure 10 illustrates the relation between the reliability sensitivity of geometric parameter errors improved and each improve-  (4) Based on the previous theoretical research, a case study example for a five-axis machine tool was used to demonstrate the effectiveness of this method.Each geometric error under certain accuracy design requirements was optimized.Accuracy grades of main feeding components of mechanical assemblies were also specified.
Despite the progress, a number of issues need to be further investigated to improve the method proposed here: the proposed method is developed on the premise that the volumetric machining accuracy is influenced only geometric errors, however, thermal errors and dynamic errors as other main errors can arise as a result of joint interface deformation between machine tool structural components and contribute to volumetric machining accuracy.As a result, the development of a method of accuracy allocation of machine tool for improving machining accuracy reliability while considering such errors would be a focus of future research.e-mails: hlaaok@163.com,zhangziling1119@126.com, chengqiang@bjut.edu.cn,45080936@qq.com;gupeihua@aliyun.com

Fig. 1 .
Fig. 1.The 3-dimension digital structure model of the NC machine tool the above equation, the values and means of the expressions such as 05 P which are subject to random variation about their nominal values, let the functional requirements be of the form Z=0 (limit state equation) which divides the parameter space into two regions, one a failure domain can be expressed as { : ( ) 0} F T g x = ≤ and the other is the safe domain.Supposing that there is a design point (the most possibility failure point)

Fig. 3 .
Fig. 3.The flowchart of method for geometric accuracy allocation of multi-axis machine tool for improving machining accuracy reliability

Fig. 4 .
Fig. 4. Initial values of the reliability sensitivity of geometric parameter errors

Fig. 9 .Fig. 10 .
Fig. 9. Tendency of the maximum values and mean values of possibility of failure with different improvements X-axis parallelism error in XY plane yb α ∆ = B, Z-axis parallelism error in ZY plane xa College of Mechanical engineering and Applied electronics Technology Beijing university of Technology, Beijing 100124, China peihua gu Department of Mechatronics engineering Shantou university, Shantou, guangdong, China 515063

Table 1
Lower body array of five-axis NC machine tool

Table 2 .
Characteristic matrices of the 5-axis NC machining center

Table 3 .
Initial values of geometric parameter errors of five-axis NC machine tool (mm) x y ∆ x z ∆ x α ∆ x β ∆ x γ ∆ y x ∆

Table 4 .
Values of the possibility of failure of machining accuracy before improvement

Table 5 .
Improved values of errors and possibility of failure in different improvements xz β ∆ yz α ∆ y α ∆ x γ ∆ x β ∆