There are alot of different methodologies have been developed to the model thermo-elastic behaviourof the machine tool in order to compensate thermal errors.
In general,methodologies can be classified into two categories i) Physicalmodelsii) PhenomenologicalmodelsPhenomenological models:-Phenomenologicalmodel constructs a relationship between input parameters (e.g. Temperature) andan output value (e.g TCP displacement). Experiments are carried out atdifferent loads and results with respect to time are observed by regressionmodel (RM). Other methods like neural networks (NN) and Fuzzy logic (FL) forcompensation also listed in phenomenological models. Physical model:-Physicalmodeling approach simulates thermally induce errors distinguished, intemperature distribution and distortions, in order to calculate TCP dislocationand enable real-time compensation. All considerations are based on physicallaws.
FEM modelsand FDM models approaches are part of physical modeling. Reduction of Thermally induceerrors:-A lot ofpeople have presented different methods to reduce thermal errors, put into netshell these methods can be classified into three categories according to thesis5946,40i) Minimizing the temperature fluctuations: for example bycooling or controlled environment condition as well as minimum heat generationii) Reducingthermal sensitivity: reducing the sensitivity of machine tool structural loopto temperature changesiii) Compensationof errors: for example by mean of mathematical modelsReducing the temperaturevariations:-Temperaturevariations can be minimized by reducing the masses of machine tool structure41,thesis594, applying cooling to a machine tool, use of oil shower, through air.By tryingto create even temperature distribution thermal error can be reduced of machinetool structure. Much lower the temperature difference will be lower the thermalerror present. Thetemperature gradient can be reduced by minimizing heat generated in elements ofmachine tool. P sekler et al 41 of thesis illustrate thermal error can bereduced by sizing down the masses of machine tool structure.
This usually appliesto construct energy efficient machine tools but also it also helps in reducingthe losses occur in machine tool. With smaller masses less energy is requiredto move them result in smaller losses and lower temperature on machinestructure.The mostcommon approach implemented widely in industry is to apply cooling to machinetool. Some approaches based on try to remove the excess heat generated inmachine tool elements.
One of the approaches 40 is to design special coolingelement for the spindle. These cooling tubes try to make us of Coanda-effect.Working principle of Coanda effect, fluid passage out from nozzle creating aprimary stream. Temperature control of air in a lithography application isshown in 42. Compensation using oil shower is used in 43 and 44. Anotheradvantage of using oil shower is that insulation from fluctuations from in roomtemperature.
Variousmethodologies to reduce thermal errors that does not directly reduce thetemperature gradient on machines but modifies it, is practice of heating andcooling elements. It can be seen during application of compensation methods tomachine (47,48,49). In order to reduce tool center point displacement keyelements of machine tool either can be heated or cooled. For special cases feeddrives are used to for reduction of angular errors on three axis machine.
Reduction of thermal sensitivity:- Other thantemperature gradients approach thermal error of machine tool can also bereduced by minimizing the sensitivity of elements to temperature changes.Meaning of this machine tool design in this way that large deformations do notoccur. This can be achieved by applying thermo-symmetrical design to machine tool. In 50, thesis 5946 boundary conditions are applied to headstockof lathe in such a way that center of axis does not move during the thermal expansion.
Thermal deformation on machine tools 51 present a methodology according tothat non-sensitive machine can be designin such way that specific directional thermal expansion do not affect thatworkpiece accuracy. Advance material forcompensation of thermal displacement:- (Thermal issues page 782)Materialoptimization can be effective in reducing the thermal errors in machine tool.Alternative materials like carbon fiber reinforced plastic (CFRP) has negativelinear expansion coefficient can be used to compensate thermal displacement ofmachine component which have positive linear expansion coefficient such asaluminum.Anotherexample thermal distortions due to local temperature gradients can be reducedby using polymer concrete in machine tool bed. Achieveable reduction is upto30%. (thermal isuue page 783) Activecompensation using adaptronic devices:- (thermal issue 783)CFRPstructure are used for active compensation of angular displacement of mainspindle of housings and heating of unidirectional carbon fibre reinforcedlaminate.In an adaptronicsystem, negative thermal expansion of CFRP- structure compensate the thermaldisplacement. Thermal sensors, controllers, and CFRP actuators make possiblecontrolled heating of CFRP laminate byheating filaments and Peltier elements.
Compensation:- thermal issue In general,Thermal displacement can be estimated in two classes of methods: Directcompensation and indirect compensation. Process chain of thermal deformation Direct methoduses touch probes to compensate error, for that machine has to be stoppedduring an operation to take measurements, the big drawback of directmethodology, ultimately productivity reduce. On otherhand indirect measurement reduce downtime by active compensation. The indirectapproach uses temperature measurement to calculate TCP displacement with helpof mathematical models. The most commonmodel used for are described below:-Method of thermalerror compensation based on linear and nonlinear regression:- Regression model is applied for error compensation, it definesa relationship between dependent and several independents values. In case ofthermal errors temperature in specific machine tool points are independentvariable and dependent variable are TCP displacement.
It is active compensationmethod which means without disturbing the machine process errors can becalculated. The hindrance with indirect compensation is the installation of measuringsystem is very costly.The drawbackof RA is selecting positions for temperature sensors if too many positions are takenit will increase the cost if few than the accuracy of the solution will becompromised.
Compensation basedon neural networks:- 8 thesisUsing Neuralnetwork approach for thermal error compensation is a common practice. Feedforwardnetworks are used for thermal error compensation, temperature probes act asinput. Neural network approximates the TCP dislocation relying on the temperatureof machine tool. Input and output layers act as input and output buffer fortemperature measurement of machine and machine thermal errors respectively.Layers in between them are called hidden layers. The working principle of theseto suppress the noise.
Each inputis multiplied by the interrelated weight. All of these weighted inputs aresummed up and combined with a threshold to find out activation level of theneuron.NET=x+thNET=summationof weighted inputs and threshold Inputssignal received by artifical neurons wieghtsPhysical models:-Compare toANN and RM a lot of others models are in practice. In 7, thesis 5946 lumped capacitance method is used tocalculate the temperature distribution of the machine tool. To do that thermalbehavior knowledge will be needed because one must know which parts of machinecan be lumped and how to apply proper boundary conditions for lumped bodies.
Aseries of temperature is used and TCP dislocation is computed by stress-freetheory and rigid body kinematics.Withadvancement in computer field and accessibility of models, e.g FEM models, newreduction procedure can be developed. Denkena et al, 42 thermal issues 785 appliedFEM to calculate thermal deformation of machine tool in steady state versusload profile. During operation TCP displacement are compensated with a linear model comparing temperaturemeasured on machine tool structure with those computed n steady state.
FDEMapproach is endorsed fot real-time compensation of machine tools 141 thermalissues 785. Unknown boundary condition, simulation-based model, Volumetric TCPdisplacement, use of thermal location and components errors as correctionvalues. A mixtureof FEM and FDM used in 76 a transient thermal analysis has beenperformed using FDM and TCP displacements has been compensated using FEM