Bogdan Supeł, Zbigniew Mikołajczyk Department for Knitting Technology and Structure of Knitted Fabrics Technical University of Łódź ul. Żeromskiego, 90-543 Łódź, Poland E-mail:katdziew@p.lodz.pl Model of the Connector for 3D Distance Knitted Fabric Fastened by Articulated Joints Abstract During the first stage of our considerations, a model of the compression process of a 3D distance knitted fabric was related to a single connector fastened to it by articulated joints, and considered as a slender rod with an assumed shape. In our physical and mathematical models, the compression of a slender, elastic rod was based on assumptions concerning the knitted fabric’s morphology, as well as the mechanical properties of the threads, particularly monofilaments placed in the internal layer of the fabric. A calculation method was developed which enables to determine the functional dependencies between the compressing force and bending deflection, as well as a method of determining curves representing the shape of the compressed rod – connector. A computer simulation of the compressed rod connecting the two outside layers of the knitted fabric, considering the variable parameters of the model, was carried out with the use of calculation algorithms elaborated by us, and MathCad program. Key words: knitted 3D fabric, distance fabric, compression process, connector, articulated joint, mathematical model, physical model, simulation. Introduction The modelling was carried out with the aim of finding the optimum procedure which would facilitate in solving the problem of 3D distance knitted fabric compression. The process of compressing such a fabric is accompanied by a number of physical, particularly mechanical phenomena, such as the bending of the connectors, which form the knitted fabric’s structure, and the mutual interaction of the friction between neighbouring connectors, among others. Designing a distance knitted fabric which would fulfil the demands of its future user is a complex procedure. Irrespective of the shape of such a fabric, described by its geometry, it should fulfil a number of usability requirements, including sufficient, mainly high strength, and resistance to the action of external forces. An analysis of literature carried out by us concerning the structure and mechanical properties of 3D knitted fabrics and the modelling of their strength behaviour led to the conclusion that the problems of modelling the compression of such fabrics have hitherto not been described. Therefore, developing a model description of the fabric’s compression process will create the possibility of designing distance fabrics of assumed elasticity without the need to manufacture a prototype. The designing process concerning geometric and technological aspects will be completed by selecting an appropriate material, and next manufacturing the fabric with the use of a suitable knitting machine without the need to carry out initial experimental stages. Such a procedure will allow to significantly shorten the designing time of 3D distance knitted fabric. Assumptions of the model The object compressed, which is a 3D distance knitted fabric, is considered as two rigid planes connected by a network of rods which form an elastic spatial trust. The shape of this trust depends on the knitted structure of the stitches, as well as on the mechanical properties of the threads which form the structure [1]. Modelling th compression process of the knitted fabric was performed during the first stage’ of the ivestigation, assuming a single connector comprising a rod joining the outside layers of the knitted fabric. a) The rod compressed we considered as placed in the xyz co-ordinate system, and to facilitate the analysis of the phenomenon, we mentally divided the rod – connector into two rods consisting of its projections on the perpendicular planes 0xz and 0xy, which are the planes of two cross-sections of the knitted fabric (Figure 1). The outside planes, p1 and p2, of the knitted fabric are composed of single loops or loops connected into loop systems. The loops consist of single systems of the outside planes p1 and p2. We consider them as basic elements of the planes’ structure with a dimension of A × B, where B is the width of the wale, and B is the height of the course. The segments do not deform, which means that A = constant and B = constant. The loops of the model b) Figure 1. 3D distance knitted fabric; a) design of the fabric, b) photograph of the crosssection of a 3D distance knitted fabric. Supeł B., Mikołajczyk Z.; Model of the Connector for 3D Distance Knitted Fabric Fastened by Articulated Joints. FIBRES & TEXTILES in Eastern Europe January / December / A 2008, Vol. 16, No. 5 (70) pp. 77-82. 77 force P acting at the midpoint of the surface of the rigid plane p1 (Figure 3). The following dependency connects the unit forces Pi with the summary force P: P = ∑ · Pi where: m – is the number of loops closed by the surface of plane p1. Figure 2. Physical model of the outside planes of the 3D knitted fabric. Figure 3. Equivalent load of the planes of the 3D distance knitted fabric. Figure 4. Fastening scheme of the physical object – a bent elastic rod. planes are connected by articulated ball joints (Figure 2). From a mechanical point of view, we consider the planes as shells which are susceptible to deflection and non-susceptible to stretching. For such a model of the planes p1 and p2 accepted by us, only small and insignificant bending moments occur (Mg ≅ 0) at the points of loop joints. Only the mutual influence of neighbouring elements, e.g. of loops, occurs due to the reaction forces (F1) in the articulated joints, which act on the surface of the shell. In the case of the compression process of 3D distance knitted fabric, in order to accept a uniformly distributed load caused by the unit forces Pi acting on the singular loop segments, we assume that this is equivalent to the impact of a singular 78 The structural model of the plane accepted by us may be related to the process of compressing the distance knitted fabric with the use of a tensile tester. In this case, the rigid dished disks connected to the measuring head of a tensile tester perform the compression process of the sample investigated. The physical model of compressing a slender, elastic rod in a single plane, 0xz or 0xy, is based on the following assumptions: the planes of the knitted fabric are displaced in the transversal direction, the loading by force P acts in a direction perpendicular to the planes of the knitted fabric, the connectors are considered as elastic rods whose structure has a spatial configuration, as well as elements fastened by an articulated joint to the outside surfaces of the knitted fabric at both ends, the connectors which belong to a single loop are joined at a single point, the planes p1 and p2 are not deformed (this is a simplifying assumption of our analysis, as the connector considered presents a whole set of connectors the fact that the knitted distance fabric analysed). In the event that such an assumption would not be accepted, irrespective of all the connector loads acting are of the same value, the connectors would be characterised by different deflections, which in turn would complicate the considerations. An analysis taking into account loads causing irregular connector deflections will be the subject of our future works and publications; in the preliminary stage of the compression process, the rods retain the assumed shape, and next, with an increasing force P1 , a transversal force S1 of the base reaction appears, such factors as temperature and humidity do not influence the proceeding of the deformation process of the knitted fabric. In a further analysis of the mathematical model based on the physical model, the connector is considered as a slender rod with an assumed shape, which is affected by the bending process. The assumed shape is obtained by accepting the thickness g0 of the knitted fabric at which the deflection value ∆g takes the zero value, and the mutual displacement y0 of the rod’s end in the direction perpendicular to the deflection, as well as the assumed length lp of the rod. The rod is considered in a free state (not loaded) and takes such a shape that its potential energy will be the smallest. Physical model of the compression process Taking into consideration all aspects related to the structure of the 3D distance knitted fabric, as well as the behaviour of this fabric under the influence of loads acting on the fabric, and on the basis of the assumptions mentioned in the previous chapter, below is presented a physical model of a slender, elastic rod of assumed shape. The general assumption is that the rods are fastened at both ends by articulated joints. As far as compressed rods are concerned, the model discussed is related to the Euler’s theory, the difference being that our assumption, confirmed empirically, considers the rod as a slender object with assumed preliminary shape; what connects is existing of additional forces S1 that occur during the compression process, which are caused by the reaction of the base i.e., the reaction of the outside layers of the knotted fabric [2-5]. Figure 4 presents a scheme of the mechanical model in the form of a rod mentally liberated from constraints; this presents the outside planes of the knitted fabric, whose action was substituted by the forces P1 and S1, and their motion is limited by both the immovable and slidable articulated joints. The following designations are marked in Figure 4: g0 – initial thickness of the knitted fabric, gi – thickness of the fabric during compression, ∆g – deflection of the knitted fabric, p1, p2 – planes in which the knitted fabric’s loops are placed, y0 – horizontal displacement of the rod’s fastening points, P1 – force compressing the bent rod, and S1 – transversal force of the base’s reaction. FIBRES & TEXTILES in Eastern Europe January / December / A 2008, Vol. 16, No. 5 (70) ed fabric,of they0knitted - horizontal of the rod’s fastening points, Sinertia reaction force ofcross-section. the transversal, sliding support, where: 1 - is the eflection fabric, displacement respectively; P1J action, - is the modulus ofDifferentiating of the bent object’s C41 = 0 ‘x’, we subsequently obtain: equation (1)and twice towards sec g compression, P – force compressing the bent rod, and E is the Young modulus, 1 – planes in which the knitted fabric’s loops S1 - isare theplaced, reaction force of the transversal, sliding support, S1 – transversal forcefastening ofEDifferentiating the reaction. J - is (1) the twice modulus of inertia of the bent object’s cross-section. ric, orizontal displacement of the rod’s - is base’s thepoints, Young modulus, and equation ‘x’,IVwe subsequently obtain: P1 y′′( x) P1 y′( x) −towards S1 ′′′ and where: ed fabric’s loops are placed, y ( x) = y0 ( x) + y1 ( x) y ( x) =of the bent object’sy cross-section. ( x) = rce compressing the bent rod, and J - is the modulus of inertia For the condition which connects the force EJ EJ Differentiating equation (1) twice towards ‘x’, we subsequently obtain: the rod’s fastening points, ansversal force of the base’s reaction. P1 y′′( x) P1 y′( x) − S1 IV , ∆g), we substitute t implicit function f(P 1 and y ( x) = where: y ( x) = y0 ( x) + y1 ( x) (1.a) y′′′( x) = rod, and Differentiating equation (1)finally twicetakes towards ‘x’, we subsequently obtain: Equation (1.a) the shape: C , C , and C determined from equaEJ EJ determined from equation system (3) into 21 31 41 Mathematical model of the EquationP(1.a) the shape: ′′( x) (3) into the condition that − S1 takes IV P ysystem y′( x)finally ’s reaction. and y ( x)2tion where: y ( xdoes y1 ( x) ) = y0 not ( x) +changing y′′′IV( x) = 1 2 = 1P compression process of a curved rod, which durin 1 EJ EJ ′( x) −finally P′′1(yx′′)( x=) 0 where: k1 describes P1 y(1.a) S1 the length of the curved rod, = y ( x ) + k y IV Equation takes the shape: 1 single connector expression: and y ( x) = where: y ( x) =EJ (1.a) y0 ( x) + y1 ( x) y′′′( x) = which does not changing during processEJ EJP1 2 IV 2 The physical model presented above general oftakes equation (1) may be we written thefollowing form of:exk1 integral = finally y ( xena) + k1 y′′( x) =The 0Equation where: ing. Finally, obtaininthe (1.a) the shape: n gi 4. gi g Figure 5. Fastening scheme of the yphysical object –+aCbent elastic rod .cos( EJ bles to formulate parametrical, functional pression Equation ( x ) = C + C x sin( k x ) + C k x ) C ( + C ( sin( k i ) − sin( k1 (i − 1) i )) + C P 11 shape: 31 1 1 21 2 41 ∑ IV 2 21 31 1 1 Equation (1.a)integral finally takes the ′′ The general of equation (1) may be written in the form of: k = y ( x ) + k y ( x ) = 0 where: n n n as well as normal differential equations i =1 ure 5. Fastening scheme of the physical object – a bent elastic integral rod . 1of equation (1) may 1 The general EJ P1 cos(connector 2+ IV(the 2 + C x + Cprocess of the fourth order, which interrelate the Mathematical model of compression of a single y x ) = C sin( k x ) C k x ) (2) in the form of:41 Determining the deflection ∆g consists in y ( x) + k111 y′′be ( x21written ) = 0 31 where: 1 k1 = 1 The general integral of equationsubstituting (1) may beeach written inofthe form load the rod with its process deformations. EJ value force P1, of: con-consists in s cal model of theofcompression of a The single connector The integration Determining the deflection ∆g constants C , C , C , and C we determine from 11 21 31 41 y (equation x) formulate = C11 +(1) C21constants xparametrical, +–Cbe sin(Ckelastic )in Crod cos( kaand xwell ) C41 into 5. Fastening scheme ofThe the physical object a31bent .31aas,parameter, rod-connector isFigure considered and integration C+sidered ,41as C we determine from the b Theaccepted general integral of may written the form 1 ,xfunctional 1of: parameter, into equation (4) The physical model presented above enables to as 11 21 equation (4) and drawin The integration constants C , C , C , and C we determine from the (2) conditions: 11 21 31 41 The integration constants C , C , C , and C we determine fromb 11 21 31 41 by us as two individual rods which are conditions: and drawing the curves of the function n yequations as normal differential order, interrelate the load of the rod ( x) = C11parametrical, +ofC21the x +fourth C k1 x) +which C41ascos( k1 x) (2) al model presented above enables to formulate functional well gi gi 31 sin( conditions: y (0) y C C y = − → + = − conditions: 0single 0 the projections of the rodmodel thephysical 0xzthe and f ( P1 ,as gi )and = ∑ (C21 + C31 ( sin(k1 i ) − sin( k1 (i − 1) Mathematical of process of11rod a+determine 5) y041→we C Crod y 41 =is −C =11−connector with its deformations. The rod-connector considered and by us two Figureequations 5.The Fastening scheme ofon the –yy,(0) a(0) bent elastic integration constants C11 , compression Cobject and from (Equation the boundary differential of the fourth order, which interrelate the the 21, C31 n n i =1 y0constants C41 = y−load → + .C =C+ −accepted (0) C C,y0410C =, − y0 3 =of −Cy11 41P 0yz planes. 0 → 11 y P1 y0 The integration , 3 11 1and 21 31 planes. 0 0yz individual rods which are the projections of the rod on the 0xz conditions: eformations. The rod-connector is considered and accepted by us as two P y P y ′′′ =y0S1 the = boundary → −C k1 cos( y ( gi )P k x)(+PC ,k41gkix1)) ==sin( l pP1k1 1y0x0 ) =P y (6) 1well Cenables ( gi determine ) =toS1formulate = 1from →− y41′′′′′′we Cg031 k133cos(31k1 functional x) +and C41 1k1f33sin( physical presented above as scheme of the physical object bent rod of . and 1 3 3 x) + C 1−y athematical model of−the compression process aand single ydifferential (0)The y0 → C–11 C yon = +the =elastic −describes ods which are projections ofamodel 0yz ) y=′′′S( 1gi=) =g1 Si 10 =→Pparametrical, sin( ) = y (connector gplanes. Ci 31→k1−cos( k k k x 41rod 0 the 0xz cos( ) sin( + C k k x C k1gxi ) = 1 g0i 1 41 1 1 i Thethe equation which 31 1 1 41 100 1 conditions Equation 3. gwhich interrelate where: n = is the number of parts into w as differential normal differential equations of the fourth order, the load of the rod where: n = 100 is the number g The equation which describes deformations of the rod influenced by the loads i y ( gi ) =P10y→ C i + C gi + C31 sin(k1 g i ) + C41 cos(k1 ggi i) = 0ofgi parts y0 influenced by deformations of theP1rod 3 the 3)=0→C 0 g 11+ C 21sin( ( ) cos( ) 0 + + = y g C k g C k g into which the interval the function is divied, which is accepted using intuition with its deformations. The rod-connector is considered and accepted by us as two ′′′ compression process of a single connector (presented )the )= = following → −C31 k1 tocos( + C41 yhas githe Sfollowing kthe k(1g→ xby 31well i +as i )+C i ) = 0 of form: e physical model enables formulate parametrical, functional 1 = above 1 x )rod yk(1gi sin( C11 C21C ntial equation which describes deformations of influenced the loads 231gsin( loads has form: 0 0g+→→ sin( ) cos( +i CkC + =C 0 = 0is acCk131 k1 41gk41i2cos( Ck1411kisgcos( k10g→ 11 2111gC21, 1 g i ) =y0 i i The integration constants C11, 21C31, i ) == i +k i )which (3) g ′′ (0) sin( ) ) − − = y C x C x 2 2 variability (2) divied, i i the calculation. The greater the value ‘n’, t rods which are order, the projections rod 0xz and 0yz1kplanes. The integration constants Cwhich , ofC=the and Cload we from41k the boundary 1 1 C 41 11, Cy21 310, → ′′interrelate (0) sin( −the Con k41the x1) −rod C normalform: differentialindividual equations of the fourth ofk311determine the owing 31the 1 2equation 41 k 1 2cos( 1 2xx))==00→ 41 ==00 2 x )sysand C41 obtained from ′′ (0) 0 sin( cos( = → − − → y C k C k C cepted using intuition by the authors as ′′ (0) 0 sin( ) cos( ) 0 0 = → − − = → = y C k k x C k k x C 31 1 1 41 1 1 41 but significantl (to cos( +′(Cx21) +giS+xC31where =)0integration gJy k1 gyi()x+) C gThe dh above enablesyE functional as41ywell , C121at, Cthe and Ctime from the equa constants C1141 1 1 41 this way 311, same 41 obtained i )formulate ′′=(conditions: its deformations. The rod-connector issin( considered and by us31 as = −C x0) → P111 yparametrical, ( xintegration )k+1accepted yi1)( xequal: (1) tem=are ,C and C obtained the equation sys The constants Ctwo 1 0relatively 11, C21giving 31, sufficient 41 accuracy to from the calcula(1) , C , C , and C obtained from the equation sy The integration constants C y (0) y C C y = − → + = − computer program. The differential which deformations of the rod influenced by the loads 2 equation 2 describes ons of the fourth order, which interrelate the load of the rod relatively equal: 11 2111 3121 41 , C , C , and C obtained from the equat The integration constants C 0 11 41 0 31 41 ividual rods which are the projections of the rod on the 0xz and 0yz planes. The integration constants C , C , C , and C we determine from the boundary 11 krelatively 21 31 C41 = 0 41 (1) y′′(0) = 0y (→ − P1 y′( x) + S1 x where x) −=Cy310 (kx1 )sin( + y1k(1xx)) − C41 k1 cos( 1 x ) = 0 → equal: tion. The greater the value ‘n’, the greater relatively equal: the following form: rod-connector ishas considered and accepted by us as two (3.a) relatively equal: = − C y conditions: P1 y0 C , C 3, C , Cand= −Cy obtained P y0 equation 11 0 3 where: the accuracy of the calculations is, but at system are The integration 11 31k1 C 0 k sin( k x )from ( gthe ) =−0xz → −C = 1 the yy′′′on Syconstants k21 x1111) += C − 41 y41 1 = 31 1 cos( 1= − y 1 i= ofequation the rod and 0yz planes. The value of ∆g, is related to point ‘ C 0 11 (0) C C y → + = − eprojections differentialwhere: which describes deformations of the rod influenced by the P1as result 0 0 11 41 0 (3) which the same time this way significantly ing g(i g loads y0 (x), and y1 (x) are the initial buckling, and the bucklingy occurring of the force relatively equal: i P EJ tg ) y0 (x), and y1 (x) are the initial buckling, 1 i 0 ′′ ′ Jy ( x ) P y ( x ) S ( x ) = y ( x ) + y ( x ) = − + E x y we read from the x-axis (see 6). Thi (1) where the following form: y0 EJ ) y PEJ P1P 1 1 1 ytg creases the(3.a) calculation time used by Figure the 0 ( gi P1 as ybuckling respectively; 1 action, 3 ych are the buckling, the as++0 yC result the =EJ +iy()gxiforce y30 of tg )1gi 0 1 ) CP y0yy(′′′g(ig 0 result ) =of 0and sin( cos( 0 → = Crod C g− Coccurring k1kgCloads k g EJ 1 (x) 0 41 C EJ tg ( 21 andinitial the buckling occurring of 11 = − 11 + 21 31 1 i + ix))= describes deformations the influenced by the ) cos( sin( ) = = → = S C k C k k 0 (5 and 6). It should be stressed that it is imp (3.b) + 1 y0gi 1 y041 EJ P 1 31 1 1 21 i computer program. (3) S1 - Pis action, the reaction force sliding espectively; the force gi C+21 =support, gi 2 of the transversal, P1 gi 2 g i 1EJ 2C21 = 2+ P 1 y ′′(0) = respectively; equation (4) as a great number of solution 0(y→ =2 0 P1EJ (1) + -the y′′( x) = −force P1 yS′( –xof)E Sis x−of )C=1 31 xsin( )and + ky11 (xx) )− C41 k1 cos(gk1i x) = g0gigi→ where where: 2 CP 41 1 y Young EJ tg g(i modulus, )yk01(transver1 xtransversal, the eaction sliding support, EJ isthe y0reaction 1 EJ11the ) =force 0→ sin( ) +and cos(C )EJ=g0i occurring +constants + CC y ( g0and C C ginitial k1 g, iC C k1i gobtained equation may beisfound graphically. T value of ∆g, which related only to point 21 31 41 i i i + CJ21-=yis (x), y (x) are the buckling, the buckling as The result of (3.b) the force EJfrom , C , and the equation system are The integration P1 1 of inertia of the bent 11 21 31 cross-section. 41 the modulus object’s sal, 0sliding support, oung modulus, and g P y EJ sec( g ) P 2 2 2 i 1 0 i ‘A’ (the point where the function charts 1 (relatively xy)′′(0)g=respectively; ere y ( x) = y0 ( x) +P1y1action, (1) y EJ sec( g ) P → −C 01 x ) = 0 → i C411 = 0 P1EJ i 0 equal: ere: of inertia E – isofthe Young 31 k1 sin( k1 x ) − C41 k1 cos( ky odulus the bentmodulus, object’s cross-section. −y gEJ EJ sec( EJ )gi EJ and ) 31 = 0C cross), we read from the x-axis (Figure 5, 0 i sec( C31 =‘x’, − sliding EJof the Sthe -CThe is the reaction force of the transversal, support, P1 EJ (3.a) 1 modulus = − y J –are is the of inertia of the bent Differentiating equation (1) twice towards we subsequently obtain: (x), and y1 (x) initial buckling, and the buckling occurring as result force , C , C , and C obtained from the equation system integration constants C − C31 = − 41 11 0 (3.c) 11 21C31 =31 P g see page 80). This isare a graphical solu1 i P1 modulus, and gi P E is the Young EJ object’s cross-section. P 1 y EJ sec( g ) action, respectively; 1 equal: 0 towards i ‘x’, ing equation (1) twicerelatively we subsequently obtain: g P EJ tion of the equation system (5 and 6). It g i 1 (3.c) EJ i ) of the the object’sCcross-section. i as result C31force =J −- is ial and the buckling of1 ybent the EJ P1impossible1 to ′′( x)force −y0 S1 tgof( ginertia (0yx0transversal, ) occurring P P1=the yy−′modulus where: - isbuckling, the reaction of sliding support, = 0 EJ (3.a) (3.b) IV EJ C should be stressed that it is 41 11 P 1 )= gi andtwice (1.a)P1sec( y ( x) = y0 ( x) +where: ywhere: +P1 y′′′( x) C=21 =gequation C41 where: =0 y ( xto)= 1 ( x) Differentiating (1) = directly g P1EJ 1 from equa1 P1 1) ) where: P1 P obtain asec( point − S1Young modulus, (isx)the P1and y′′(igxEJ EJ C41 = 0 C41 = 0 where: 2 i )EJ sec( gi i (gEJ = IV i, P1)g P1cos( gi ) sec( ) = g 1 and where: (1.a) i y x y x y x ( ) = ( ) + ( ) wards ‘x’, we subsequently obtain: = y x ( ) i tg ( g Differentiating equation (1) twice towards ‘x’, we subsequently obtain: he transversal, sliding support, EJ y EJ ) 0 1 g cos( ) P EJ of i 0 isEJthe modulus of inertia EJ of ythe bent object’s cross-section. tion (4) as a great number of1EJ P i solutions 1 EJ (3.b) g cos( ) EJ EJ 0 g cos( ) P 1 i (3.d) where: C41 = 0 C21 = + 1 d EJ deformation (3.d) connects this equation The solution ofi this sec( gi For ) = the condition which EJ the exists. force P the ∆g in Equation (1.a) 1 with gi finally2 takes PP11 ) the shape: For the EJ condition which connects the force P with the deformation ∆g in the for P 1 EJ sec( g ′ ′′ − ( ) P y x P y x S ( ) 1 g fferentiating the bent object’s cross-section. 0 i equation may be found only graphically. IV 1 1i 1 For the condition which connects the force P with the deformation ∆g in the fo g cos( ) equation (1) twice towards ‘x’, we subsequently obtain: 1 P (3.c) For the condition which connects the force P with the deformation ∆g in , ∆g), we substitute the integration constants C ,C implicit function f(P i EJ ′′′(C and where: (1.a) 1 y ( x)f(P y ( x) + y11( x) substitute the integration = EJ .a) finally takes the EJ y (2x) = 1 implicit constants C11, C21, C1131, function k1 = y IVy shape: ( xx))31+==k−12 y′′EJ ( x) = 0 where: 1,0 ∆g), we The reasonthe forintegration this the is also that a symEJ , ∆g), we substitute constants C , C , C implicit function f(P P determined from equation system (3) into condition that describes 1 11 21 31t2, substitute the integration constants C11 ,C implicit function f(P , ∆g), we P1 1 EJforce P1 2 condition 2 gwhich determined from equation (3)form into theancondition that and describes lengt For kwe the connects theFor Pcondition deformation ∆g in the of the which connects 1system the metrical deflection line exists that the i ′′ obtain: 1 with the twice ‘x’, subsequently = y′′( x)Ptowards = 0 where: y EJ sec( g ) ′ − ( ) P y x y x S ( ) determined from equation system (3) into the condition that describes the leng 0 i 1 EJ IV curved rod, which does not changing during processing. Finally, we obta exists ‘different’ buckling shapes occur, which will(3.c) be 1 The general 1 from equation system (3) into the condition that describes th of equation (1) may bexwith in the form of: EJ and anddeflection where: (1.a) y ( xforce ydetermined )so-called = ycurved )written +integration ( x) deformation (1.a)that ( x) = 1 = line EJ( x)integral rod, which does∆g not processing. Finally, in the ‘different’ buckling shapeswe oc-obtain the fo 0P(1the 1the we substitute constants Cchanging , C21so-called , Cduring , and C implicit Cyfunction 1, ∆g), 11 31 41 Equation finally takes the shape: 31 = −(1.a)f(P curved rod, which does 1not changing during processing. Finally, we obtain the f EJ EJ expression: discussed in the subsequent parts of this article. curved rod, which does not changing during processing. Finally, we obtain P (3.d) where: P C = 0 1 y ( x ) = C + C x + C sin( k x ) + C cos( k x ) (2) l integral (1)41 11may be the form41 of:P form of1 an implicit cur, whichof will expression: 21 written 311 in system 1 sec( gi function )that = f(Pdescribes 1, ∆g), we the determined from (3) into the condition length thebe discussed in the subseP1 y′′(of x) equation gequation 2 IV y ( x ) = y 2 ( x )i + y ( x ) 1 expression: (1.a)theexpression: EJ (2) PC EJ ′′ integration constants quent parts of this article. = y ( x ) + k ( x ) = 0yx 10 where: k1 substitute 1 11, +( xC) 21=x +EJ C31 sin(where: k x ) + C cos( k ) 1 curved rod, does not changing during processing. Finally, the following cos( gi we obtain ) 1 41 1 n uation (1.a) finally takes thewhich shape: EJ g41 g i boundary gi g 2 EJ g i The integration constants C , C C we determine from the n 21, C31, and P i 1 11 where: C = 0 1 g g g g gik1 (i2 − 1)gi i ))) + C31 (i sin(k1 i ) − sin(ik1 (i − 1) )) + C(3.d) ) − cos( expression: 41 n i g (C21 ) = 41 (icos( k1 i sec( n∑ i in g g g g g gni) 2 2 = C ( + C ( sin( k i ) − sin( k ( i − 1) )) + C ( cos( k i ) − cos( k ( i − 1) ))) + ( TheFor general integral of equation (1) may be written the form of: P 2 n n n n g g g g g ∑ i i i i i 2 V 2 21 31 1 1 41 1 1 conditions: the which connects the force of EJ i the (C21∑ni =1P+ ( sin( i(deformation ) k−1sin( −∆g 1)kn1 in +1)C41iform ())cos( i cos( − 1)kn1 (i))) k1 condition = 1 ) + k1 y′′( x) = 0 where: nsin( nan) k−1cos( (1CCwith + Ck31 iP1 )ik)1 −(isin( (i))−the + Ck41 i i k)1−(icos( − 1)+ (ni ))))2 += 1cos( 1( i∑ =1 he( xshape: 2131 g i n integration nconstants n y ( ximplicit )y=(0) C11= +−function C x +CC x)y+0we C41 substitute cos( i =1k1 x ) i =1the n n n nn y021EJ C , k= → +sin( 31 1− ∆g), Cg11,nC21, Cn31, (2) and Cn41 n gi g11i f(P141 gi gi gEJ P1 2 i i 2 2 (4) +the C31 ((1) sin(may k1 iP ybe ) −equation sin( k1 (i −in 1) + the C(3) cos( k1the iP with )condition − cos( k1the (P i −that 1) ))) + ( ∆g ) the = lthe e kgeneral integral of(Cequation written the))form deflection ∆g consists inofsubstituting each value of forc ∑ 21 41 (of: determined from system intoDetermining describes length For condition force the deformation form ofthe an 1 = 1nthe 3 n 1n 0 which connects 1 y0 n n in ∆g consists in psubstituting each value of force P1, con i =1 cos(kDetermining sin( =CS1 cos( = → −Cnot = Finally, ynk′′′(xg)i )+rod, x) + CDetermining k13aprocessing. k1deflection x) the the deflection ∆g consists in substituting each value 31 k1changing 1Determining 41as parameter, into equation (4) and drawing the curves of of theforce function: 1, co x) = C11EJ + C21 x + C31 sin( k x ) (2) curved which does during we obtain the following deflection ∆g consists in substituting each value of Pforce 1 41 1 f(P1, ∆g), we substitute the integration C11drawing , C21, C31 , and C41 of the function: implicit function (3) gconstants as a parameter, inton equation (4) and the curves i (3) the ion (1) may be writtenexpression: in the form of:gi as a parameter, into equation (4) and drawing the curves of the function: g g g g as a parameter, into equation (4) and drawing the curves of function: i i i determined from system (3) C intof n(the that the length of the g31i( sin(Pk11 ,i considered g k (i − 1) gi k1 (2i − 1)g gcondition + force Cdescribes ) − sin( )) + C41g(i cos( k1 i i ) − cos( deflection ∆g consists in substituting of ∑ i value i )g=g nP1 ,k 0(31C(21sin( =0→ y ( gi ) the C11 + Cequation x) + C41 cos(k1 xDetermining ) 21 g i + C31 sin( k1 gf(2) 1 21 i n)gi == (i )P1+ , gi )41= cos( (each C C − sin(gk1 n(i − 1) gi i))1 + C41g( cos( n k1 i gi)we n k1 i gi) − cos(gki1n(i − 1) g))) 2+ ( gg ∑ gsin( i+ i 1 i i i curved rod, which does not changing during processing. Finally, obtain the following f ( P , g ) = ( C + C ( k i ) − sin( k ( i − 1) )) + C ( cos( k i ) − cos( k ( i − 1) n the n n n ∑ ( ( C + C ( sin( k i ) − sin( k ( i − 1) )) + C ( cos( k i ) − cos( kn1 ())) i − 1)+ (ni i f curves 1 the 31 1 1 41 1 1 iP =11 , g i ) =21 as a parameter, into equation (4) and drawing of function: 2 2 ∑ 21 31 1 1 41 1 n ′′ n n n n n Figure 6. Determination of one of the points (P , g ). = g0 → −C31 k1 sin( y (0) x)i ==1 0f→ n n g k x) − C k1 cos( g k and i =C 1 g = 0 =n l pk (i −11)g gi i )))n2 +g( gi ) 2 = l n(4) expression: n (C21 i g+i C31 ( sin( k1 i gi )i 1− sin( k1 (41i −and 1) gii ))f +(1C ( cos( cos( =(k=1P P li1lpg,41i g)i )−i )−cos( 2 1 i i 2 (5) p 141,(,g i ))k f ( P1 , gi )∑ = (C21 k1 i n ) − C sin( k ( i − 1) )) + C cos( i k ( i − 1) ))) + ( ) ( f P g n + C31 ( sin( n n n n and 1 41 1 1 ( , ) = f P g l i =1 ∑ integration , C , C , and C obtained from the equation system are The constants 1 i 41n1 =np100 11 21 n 31 and i where: isp the number ofnparts into which the interval of the functio n n n i =1 where: ni, P =1100 the number of Pparts into which the interval of the function variab )100 of is the dependency Irrespective Inrelatively nthis way we obtain one of the points 1 g= f(∆g). gi equal: gi g i (gwhere: g gwhich 2 parts where: n = is the number of parts theby interval of (4) theasof function varia i i i into is divied, which is accepted using the giving suffic n k=is1 i100 iscos( thekusing number ofinto theauthors interval the function (4) ( sin(dependency k1 i ) − sin( kP −=1)divied, )) + C ( cos( )− i − 1) intuition ))) + ( intuition ) 2(6) = l which , g(Ci )21 = l+p C31the Pdetermining and fof(∑ p authors 1 (i is 41is 1 (draw which accepted by the giving sufficient acc 1 f(∆g), it also possible to deflection lines for as 1 (3.a) Determining the deflection , considered ∆g consists substituting each value of force P Ci =111 = − y0 n n n inisthe n n n 1 is divied, which is accepted using intuition by the authors as giving sufficient calculation. greaterusing the value ‘n’, by thethe greater theasaccuracy of acc the divied, which isThe accepted intuition authors giving suffici the calculation. The greater the value ‘n’, the greater the accuracy of the calcula bent rods by substituting the points (g , P ) obtained into the rod deflection equation where:the n = 100 is the number of parts into which the interval of the function variability (2) i 1the curves as a parameter, into equation (4) and of the function: thedrawing calculation. Thesame greater the value the‘n’, greater the accuracy of the of calcul butcalculation. at the time this way‘n’, significantly increases the calculation ti P1 the The greater the value the greater the accuracy the but atthe theauthors same time thisg way significantly calculation time used yn0 way EJ tg (we ggi obtain ) (2). Iny thisis a∆g rod deflection related to the point (gi, Pincreases )i 2in(3.b) the the is divied, which accepted using intuition as giving sufficient accuracy 1increases gi consists gequation gsignificantly gto butby at the time this way significantly the calculation time use i i same i value i P21, considered Determining the deflection in substituting each of force computer program. (5) EJ but at the same time this way increases the calculation tim 0) = f ( P , g ( C + C ( sin( k i ) − sin( k ( i − 1) )) + C ( cos( k i ) − cos( k ( i − 1) ))) + ( ) (5) C 1= i +The ∑ 21 n the 31 1 1 greater 1 computer program. following form: the calculation. value thedrawing the41curves accuracy of function: the1 calculations is, n‘n’, nprogram. n n computer gi i =1 greater as21a parameter, and the ofnthe P1 equation (4) computer program. 2 into g but at and the same way significantly increases the calculation time used by the n f ( P ,time g )i =this lEJ g g gi value of ∆g, gwhich g to2 point gi 2 ‘A’ (6) The is (related point where the functi i (5)(the P1 , gi ) =1 ∑ i (C21 pi + C31 ( sin(k1 iP i ) −The sin( kvalue + CP41which ( cos( k1 iis related ) − cos( k1to i −point 1) i ))) ) point ‘A’+ ((the where the function chart computerf (program. 1 (i − 1) of))∆g, 1n 1 which The value of ∆g, is related to point ‘A’ (the point where thesolution function n n n n n we read from the x-axis (see Figure 6). This is a graphical ofchar the P i =1 The value of ∆g, which is related to point ‘A’ (the point where theequation functio )interval y0 EJ tg ( gi of )parts yread gi the 1 number where: yn0 = into which the the function variability (2) solution 0 EJ sec( from (see 6). This is a graphical of the EJ100 sec( ) y0 gisi the P1 ofFigure EJ ) we EJx-axis (7) we read from the x-axis (see Figure 6). This is a graphical solution of the equatio (3.c) (5 and 6). It should be stressed that it is impossible to obtain a point (g (7) EJ ( ) ( sin( ) = − + + − y x y x x we read thestressed x-axis (see Figure 6). This is aobtain graphical solution thei, eP ( P01which , g i ) =isisl paccepted isand using intuition thefrom authors as giving accuracy to C31divied, = −f∆g, and(the 6).by It should be thatsufficient it is impossible to(6) a point (gi, P1of ) direc The value of which point(5 ‘A’ point where the function charts cross), gi P related EJ Pto P 2 1 1 (5 and 6). It should be stressed that it is impossible to obtain a point (g , P ) (4)should as aaccuracy great number of of this exists. The i 1 (gdirec (5equation andas6).the be stressed that itsolutions is to equation obtain point 1 gnumber i, P1 calculation. The theofvalue ‘n’, the the ofsolutions the calculations is, i i(4) greater g100 equation a Itsolution great number ofimpossible this equation exists.aThe solution where: n =x-axis is thegreater parts into which interval ofthe theof function variability (2) i we readthe from the (see Figure This is a ggraphical of equation system EJ 6). EJ equation (4) as a great number of solutions of this equation exists. The solutio EJ equation may be foundnumber only time graphically. The reason for this is also tha equation (4) as a great of solutions of this equation exists. The but at the same time this way significantly increases the calculation used by the equation be authors found only reason to for this is also that a syms is divied, which is accepted intuitionmay by as giving accuracy (5 and 6). It should be stressed that using it is impossible to the obtain a point (ggraphically. , P1)sufficient directlyThe from igraphically. equation may be found only The reason for this is also that a sym P 1 equation may be found only graphically. The reason for this is also that (3.d) where: C = 0 1 computer Equations 3, 4, and 7. 41 5calculation. the greater the value of ‘n’,this thegequation accuracy of theconsists calculations sec( ) = the equation (4)of aspoints aprogram. great number of solutions exists. The solution of this of theis, i greater way The row (PThe 1, ∆g = g0 – gi) determined in the EJ above-described P1 but at the same time this way significantly increases the calculation time used by the ) equation may characteristic be found onlyP graphically. Thecompressed reason for thiscos( isgialso that a symmetrical frelated (∆g) (Figure 7).charts cross), FIBRES &mechanical TEXTILES in Eastern Europe Vol. 16, of No.to 5the (70) EJ the 79 The value ofJanuary ∆g,/ December which1/ A=is2008, point ‘A’ (the monofilament point where function computer program. Forread the from condition which(see connects the deformation the form of an we the x-axis Figurethe 6).force This P is1 awith graphical solution of∆g theinequation system , ∆g), we substitute the integration constants C , C , C , and C41 implicit function f(P (5 and 6). It should be stressed that it is impossible to obtain a point (g , P ) directly from 1 11 21 31 i 1 charts cross), The value of ∆g, which is related to point ‘A’ (the point where the function Simulations of the compression process on the basis of the mathematical model Figure 5. Determination of one of the points (P1, gi). Figure 6. Mechanical characteristic P1 = f (∆g) of the compressed monofilament. Figure 7. Dependencies P1 = f (∆g) for various values of the rod’s length: 1-for lp = 12 mm, 2-for lp = 15 mm, 3-for lp = 20 mm, 4-for lp = 30 mm, 5-for lp = 50 mm. Figure 8. Dependencies P1 = f (∆g) for various values of rod rigidity: 1-for EJ = 20,0 cN mm2, 2-for EJ = 8,8 cN mm2, 3-for EJ = 4,4 cN mm2, 4-for EJ = 2,2 cN mm2, 5-for EJ = 1,1 cN mm2. In this way we obtain one of the points (gi, P1) of the dependency P1 = f(∆g). Irrespective of determining the dependency P1 = f(∆g), it is also possible to draw deflection lines for the bent rods by substituting the points (gi, P1) obtained into the rod deflection equation (2). In this way we obtain a rod deflection equation related to the point (gi, P1) in the following form (Equation 7). accepted, we can obtain a general characteristic of the whole spatial knitted fabric. The force compressing the whole knitted fabric should only be divided by the number of existing connectors, and for the force thus obtained, the mechanical characteristic P1 = f (∆g) would be designed. Data from comparable results attained by the mathematical simulation and by experimental tests obtained with the use of a measuring stand specially designed by us are not included in this article, as the stand is actually used later in the investigation; however they will be the subject of the next publication. The row of points (P1, ∆g = g0 – gi) determined in the above-described way consists of the mechanical characteristic P1 = f (∆g) of the compressed monofilament (Figure 6). The mechanical characteristic was calculated and drawn for the following parameter values: knitted fabric thickness g = 10 mm, rod rigidity EJ = 4.4 cNmm2, rod length lp = 15 mm, mutual displacement of the monofilament fastenings in the planes of the knitted fabric y0 = 2 mm. The curve has a strongly non-linear character with a large increase in the bending force P1 at the greatest rod deflections. The mechanical characteristic P1 = f (∆g) has a similar character to the one obtained empirically [10]. Notwithstanding that the mathematical analysis carried out by us concerns a single monofilament, and thanks to the assumptions for the physical model 80 For our simulation we accepted that the thickness of the compressed knitted fabric would be g0 = 10 mm, and the rodconnector would consist of a polyamide monofilament (JPA) of d = 0.14 mm. The tensile elasticity modulus was assessed empirically with the use of tensile tester from Hounsfield. The value of the Young modulus characterises the material of the rod tested . An increase in the Young modulus makes material more rigid, which means that it is not so susceptible to deformation, whereas a decrease in rod (polyamide monofilament) diameter results in an increase in the modulus of inertia J of the cross-section of the bent object, and consequently an increase in rod rigidity. Aiming at the simulation of the compression process of a monofilament placed in the internal layer of a spatial knitted fabric, two calculation algorithms were elaborated, one for calculating the compression characteristic P1 = f (∆g), and the second for determining the curves representing the shape of the compressed rod. The calculations were carried out with the use of the MathCad computer program. As a result of the computer simulation, the deflection dependencies were determined as a function of the rod length lp and rod rigidity EJ, illustrating the influence of these parameters on the values of forces and deflections (Figures 7 and 8). Figure 9 presents the curves P1 = f (∆g) for different values of the rod length. Only small differences are visible in the shape of the particular dependencies. This results from the fact that the functions representing the deflection lines of the rod in the model described have a similar shape, notwithstanding relatively great differences in the lengths (significantly, the curvatures of the deflections do not mutually differ). The examples shown in Figure 7 indicate that for various lengths lp and for the same value of the force P1 different deflections ∆g occur, or the opposite, formulated for the same deflection ∆g different forces P1 ,is also indicated. For example, if the deflection of a rod of length lp = 12.5 mm equals ∆g = 7.5 mm, then the bending force is P1 ≅ 7.72 cN, whereas for a rod of length lp = 50 mm the force is P1 ≅ 6,71 cN. Thus we can conclude that the mutual relations between the force and deflection depend on the length of the rod. On the other hand, Figure 8 presents the dependencies P1 = f (∆g) for various rigidity values EJ. In this figure it is clearly visible that for various rigidities EJ of the rod and the same force P1, different deflections ∆g occur, or in the opposite case, for the same deflection ∆g, different forces P1 are indicated. For example, if for a rod with a rigidity of EJ = 1.1 cNmm2 the deflection equals ∆g = 7mm, then the bending force is P1 ≅ 1.2 cN, whereas for a rod with a rigidity of EJ = 20 cNmm2, the bending force will be P1 ≅ 1.2 cN. Thus means that the rigidity of the rod increases nearly 18 times, causing the resistance to mechanical loading or load capacity of the rod to increase 30 times . FIBRES & TEXTILES in Eastern Europe January / December / A 2008, Vol. 16, No. 5 (70) Thanks to the analysis carried out, it is visible that changes in the rigidity of the rod-connector have an significant influence on the character of the dependencies P1 = f (∆g), as they cause a low number of changes in the shape of the dependencies; a significant displacement to the left of the chart results in the same forces P1 causing a much smaller deflection ∆g. Deflection x, mm 8 6 4 2 10 8 Buckling y,mm 6 4 2 0 Figure 9. Simulation of the rod’s deflection lines under the impact of force P1, obtained using the mathematical program. Figure 10. Families of curves representing the shape of monofilaments with the same deflections ∆g, and differentiated by parameter y0. Figure 11. Different shapes of rod deflection caused by the action of force P1. A simulation of the shape of monofilaments consisting of 3D knitted fabric, which are under the influence of the compression force P1, was also carried out. The first of the simulations performed illustrates the shape of a monofilament under the influence of force P1. From observation of the connector’s geometry, as well as the mathematical analysis performed, it can be concluded that the rod representing the monofilament takes the shape of a fragment of a sinusoid. The shape of this sine curve depends on the value of the force P1. With an increase in the force value, an increase in the amplitude of the curve also occurs, but the it tends to be narrower. The cause of P1 increasing is, from a physical point of view, a decrease in the curvature radius of the sinusoid’s apex. Figure 9 presents a broad spectrum of deflections caused by the action of force P1. The next simulation was carried out with the aim of finding the influence of the distance knitted fabric’s geometry on the value of the compression force P1. It is important here to determine the influence of this force on the mutual displacement of the connector’s (monofilament’s) ends in the direction of the y-axis (parameter y0 is designated in Figure 4). Parameter y0 has three different values: for the symmetrical system y0 ≈ 0 mm, and for the asymmetrical system y0 = ± 2 mm. Figure 10 presents 4 families of curves representing the shape of the monofilament with the same deflections ∆g but differentiated by the parameter y0. lines 1 – force P1 = 0.5 cN for yo = 2 mm, P1 = 0.4933 cN for y0 = 0 mm and P1 = 0.4862 cN for y0 = -2 mm, lines 2 – force P1 = 1.2 cN for y = 2 mm, P1 = 1.193 cN for y0 = 0 mm and P1 =1.043 cN for y0 = -2 mm, lines 3 – force P1 = 6 cN for y0 = 2 mm, P1 =5.327 cN for y0 = 0 mm and P1 = 4.74 cN for y0 = -2 mm, lines 4 – force P1 =100 cN for y0 = 2 mm, P1 =85.96 cN for y0 = 0 mm and P1 = 74.41 cN for y0 = -2 mm, The differences in the bending forces, which cause the same deflections, result from the various geometries of the rods (parameter y0). The rods are differentiated by shape (different functions describe the shape of the curves), and with them are connected different curvature radii FIBRES & TEXTILES in Eastern Europe January / December / A 2008, Vol. 16, No. 5 (70) of these lines, which finally leads to the situation that different forces cause the same deflection. The differences in the force P1 values change with an increase in these values:beginning with small differences of P1 ,causing deflections of ∆g = 0.615 mm, to relatively high differences in P1, which bring about deflections of e.g. ∆g = 9.287 mm. This directly results from the dependency P1 = f (∆g); as for greater deflections the curve is steeper. The analysis also allows to state that the greatest value of P1 related to the same deflection falls when the value of parameter y0 = 2 mm. From the point of view of geometry, this can be explained by the curvature radius of the sinusoid apex related to the shape of the connector, which reaches its smallest value at this parameter value. The smaller the curvature radius of the bent rod fragment, the higher the bending tensions are in this fragment. On the other hand, the force values, which cause this deflection, depend only on the value of this spectacular (smallest) curvature radius. The third simulation concerns the influence of the monofilament’s shape on the force compressing it. This is a problem of the so-called buckling form that takes place. This phenomenon consist in them that at equal deflections ∆g of the same rod with the same geometry of end fastenings some different values of force P1 may be created. Figure 11 presents the following different forms of rod deflection created by the impact of force P1: 1-3 lines of the rod deflection under the impact of bending force P1 with values of – 1.315 cN, 5.25 cN, amd 11.79 cN, causing deflections ∆g of 4 mm; 2-3 lines of the rod deflection under the impact of bending force P1 with values of – 3.065 cN, 12.25 cN, amd 27.6 cN, causing deflections ∆g of 6 mm; 3-3 lines of the rod deflection under impact of bending force P1 with values of –12.7 cN, 50,07 cN, amd 116 cN, causing deflections ∆g of 8 mm. From observation of the connector’s geometry, as well as from the mathematical analysis, it can be concluded the force causing a deflection in the case of the first buckling form is more than four times smaller than that in the case of the second buckling form, and eight times smaller than for the third case. 81 The practical application importance of the above-described simulation results from the fact that the rod-connector, being an element of the knitted fabric, is not isolated. The neighbouring connectors interact by contact, and the greater their number in the given volume of the knitted fabric, the greater the probability of such interaction is. Also the increase in deflection ∆g causes a decrease in the volume of compressed rods. This, in turn, led to the mentioned increase in the probability of interaction between neighbouring monofilaments. Contact between neighbouring monofilaments causes the occurrence of higher forms of buckling, and this, in turn, increases the value of the compression force P1, causing the same deflection ∆g. From the curves in Figure 11, it is visible that the trajectories related to equal deflections ∆g are mutually tangential , whereas the beginnings origins of the trajectories are tangential only for odd or even numbers. Summary This article presents a physical and mathematical model of the process of compressing an elastic, slender rod of assumed shape fastened both sides by articulated joints. The model presented is related to the process of compressing a 3D distance knitted fabric, which we consider as a spatial trust. 1. The physical (mathematical) model is based on the following assumptions: the knitted fabric planes p1 and p2 are displaced in a perpendicular direction without deformation, the connectors are considered as elastic rods with a structure of spatial configuration , and there are elements fastening both of the sides to the outside planes of the knitted fabric by articulated joints. On the basis of the physical model, a mathematical model was developed which is a differential normal equation of the 4th order. The solution of this equation at fulfilled boundary conditions and the length condition of the rod (lp = constant) is the deflection line of the compressed rod. The boundary conditions (integration coefficients C12, C21, C31, and C41) determine the physical (mathematical) model accepted. As a result of the mathematical analysis carried out, we obtained the following two equations: an equation determining the set of points comprising the mechanical 82 characteristic of the compressed rod, and an equation determining the shape of the compressed rod. 2. A computer simulation was carried out of the compression process of a model connector of the knitted fabric which included: elaboration of the mechanical characteristic of the compressed monofilament P1 = f (∆g), simulation of the influence of rod length lp, and such structural parameters as rigidity EJ. the system’s characteristic, the influence of the fastening geometry of a singular monofilament on the value of force P1, causing the monofilament’s deformation, Influence of different forms of buckling of the compressed monofilament which may occur on the value of deflection ∆g at the impact of compression force P1. The simulation carried out demonstrated that: The mechanical characteristic obtained is a strong non-linear curve; at this stage of the investigation we did not find a mathematical equation of this curve which would connect the parameters of the system together. The change in parameter lp causes small changes in the mechanical characteristic P1 = f (∆g); for example, for the compression force P1 = 5 cN, the deflection difference ∆g between the lengths lp = 50 mm and lp = 12.5 mm amounts to 0.2 mm. A change in the parameter EJ causes great changes in the mechanical characteristic P1 = f (∆g); for example, for the compression force P1 = 5 cN between rigidities EJ = 20 cN and EJ – 1.1 cN, the deflection changes from ∆g = 5.7 mm to ∆g = 1.6 mm, respectively. A change in the parameter y0 (mutual displacement of the connector’s fastening) influences the value of force P1, e.g. from 3% of the P1 value for a deflection ∆g = 0.6 mm to 34% of P1 for ∆g = 9.3 mm. For the third form of the buckling of the bent rod, the force P1 is more than 200% greater in comparison with its value for the second buckling form, and more than 800% greater than for the first form for the same deflection ∆g (these values differ slightly for different deflection values). Conclusions The investigations carried out should be treated as an introduction to further research aimed at determining an optimum calculation model for 3D distance knitted fabrics. This involves determining the mechanical properties of such knitted fabrics, but firstly their susceptibility or resistance to compressing must be evaluated. In further parts of our research, we will design and build an empirical testing model with the aim of verifying the correctness of the mathematical models already elaborated, as well as future, new models. The next stage will be the adaptation of the calculation mechanism developed to solve real problems occurring while designing 3D distance knitted fabrics., e.g. a case of spherical loading will also be considered. This investigation carried out by us and presented herein of an element of 3D knitted fabric in the form of a single connector, which means the elaboration of the mechanical characteristic of the compression process for this connector and the determination of its shape will be the basis for further research – more general and dedicated to designing 3D distance knitted fabrics References 1. K. Kopias, „Technologia dzianin kolumienkowych”, ed. WNT, Warszawa 1986. 2. W. Żurek, K. Kopias, „Struktura płaskich wyrobów włókienniczych”, ed. WNT, Warszawa 1983. 3. M. E. Niezgodziński, T. Niezgodziński, „Wytrzymałość materiałów”, ed. PWN, Warszawa 2000. 4. J. Misiak, „Stateczność konstrukcji prętowych”, ed. PWN, Warszawa 1990. 5. A. Bodnar, „Wytrzymałość materiałów”, ed. By Technical University of Krakow, Kraków 2004. 6. R. Mosurski, „Mathematica”, ed. University Scientific Editors, Kraków 2001. 7. J. Misiak, „Obliczenia konstrukcji prętowych”, ed. PWN, Warszawa 1993. 8. A. Biegus, „Nośność graniczna stalowych konstrukcji prętowych”, ed. PWN, Warszawa-Wrocław 1997. 9. W. Korliński, „Technologia dzianin rządkowych”, ed. WNT, Warszawa 1989. 10. J. Grabowski, „Dzianiny odległościowe w materacach”, Przegląd – WOS 10/2006. Received 20.08.2007 Reviewed 30.10.2007 FIBRES & TEXTILES in Eastern Europe January / December / A 2008, Vol. 16, No. 5 (70)