Development of a control system for Vibration Assisted Grinding Heisum Ewad, Supervisors: Chen X, Yu D and Batako ADL Advanced Manufacturing Technology Research Laboratory, GERI, Liverpool John Moores University

Outlines Introduction to Grinding Project Aims &Objectives Problem Outline Control System Experimental Setup Data Acquisition System Identification System Modelling Controller System Calibration Conclusion Plan for Further Work

Introduction to Grinding is an operation in which material is removed from the workpiece by a powered wheel. Types of grinding Centerless grinding Cylindrical grinding Internal grinding Surface grinding

Introduction to Grinding Grinding is a popular method of material removal Construction Aviation & Automotive Optical lenses Giant telescope mirrors

Problem Outline One of the main problems in grinding is the growing vibration (chatter) between wheel and workpiece during the process. Regenerative vibration, a type of chatter that starts as a slip-slide interaction between wheel and workpiece. However, a possible method to prevent this type of chatter is to apply a periodic disengagement of the wheel from the workpiece and periodic variation of the work- speed. These vibrations can be applied to the work-piece by devices such as piezo-electric actuators.

Why Vibration In Grinding? Reduction of cutting forces. Better coolant delivery over the entire contact zone. Better heat removal from the grinding zone. Oscillation allows the grains to cut with more than one edge. Oscillation reduces the load per grain therefore, reduces the wheel wear. Better surface finish due to lapping / polishing effect Shelf-Sharpening Process of the Wheel

Project Aim & Objectives Aims Design, model, build and control a two dimensional piezo actuator for the grinding/machining operation. Objectives Initial Development of a controller for one axis Implement control strategy within Matlab & Labview Design two dimensional controller for two axis in Matlab Design a Phase shifter to achieve oriented elliptical oscillation Implement control strategy within Matlab & Labview to control the jigs Experimental work

Control System : Theory The goal of controller is to make the system performs in a pre-defined and desired manner. Control systems are basically classified as: Open Loop and Closed Loop Open-loop control system Closed-loop control system

Control System : Theory of PID PID stands for Proportional, Integral and Derivative. Controllers are designed to eliminate the need for continuous operation attention. KP to decrease the peak time. KI to eliminate the steady-state error. KD to reduce the overshoot and settling time. Pout: Proportional output I out: Integral output D out: Derivative output e: Error τ: Time in the past contributing to the integral response

Modelling of the System System motion A mathematical was derived from the actual oscillating jig designed for the preliminary investigation. The equations of motion of the system are as follows:

Control System: Closed-Loop Preliminary Results Control System: Closed-Loop Controller Response

Experimental Setup System Set up in Closed Loop Control

Data Logging In Labview Front panel of data Acquisition In Labview

Controller Design Flow Chart

Parametric System Identification The main procedure to parametric identification are displayed in the flow chart in the next Slides. It starts with designing the experimental setup, then collecting the I/O data and pre-processing. A model structure is defined according to the selected design. The selected model structure parameters are then estimated and the model is validated. An iteration is required to go back to modify the methods and techniques during the identification until an acceptable model obtained.

System Identification Flow Chart illustrating the system Identification Process

Full system test The first element of this design of the experimental setup is the selection of the input and output signals according to the purpose of the system identification and control process. The setup should allow the designer to repeat a certain perturbation signal if necessary and to record the I/O data with a sensible time rate. Data Result From Labview Data Result From Labview

Modelling Strategies Modelling strategies are classified into three categories according to the approaches used: White box Model Gray box Model Black box Model Why I used the black box? This approach is wholly dependent on the use of I/O data collected from the physical system in real experiments. System identification, in this case, is the behavioural approach for developing a model without requirement of physical understanding of the process. The outcome of the system identification process is a model representing the original dynamic system. For the black box model, the structure is defined first before calculating the model parameters.

Data Result By using the Black Box system identification and selecting the ARX (Auto Regressive eXogenous) structure, which deals with linear systems. Model Candidate System Validation

Controller The parameters Space Method allows defining the Value of the Ki and Kp For the PI controller the maximum value is selected from the parameter space to ensure that the entire response is represented.

Equipments – Abwood 5025 machine Machine Specification Spindle motor power 1.5KW Spindle speed 5000 RPM Cross traverse of head Resolution 260 mm 10 µm Longitudinal traverse 530 mm Vertical traverse of head 350 mm 1 µm Automatic feed Pneumatic control x, y, Z Axis Maximum wheel size 400mm*25mm

Calibration Result Dynamometer: Measuring Range: 10 kN Z axis X axis Y axis

Calibration Result Displacement(mm) Dial test Indicator For the calibration process, the sensor was clamped in a support with the sensor tip in contact with the work holder. Different gauges were placed between the sensor and the work holder and the output voltage was recorded. The relationship between the output voltage and the sensor’s displacement was established. Sensor gauge Slip gauge

Actual Experimental Configuration

Conclusions Literature review has been carried out but still on-going A mathematical model of the system motion was derived The 3axis kistler dynamometer has been calibrated for force measurement The displacement sensor has been calibrated. The Data Acquisition system was built in Lab view Initial data from the stand alone rig were recorded using Lab view An initial closed loop control for one axis was developed in Matlab/simulink and required fine tuning in actual grinding The initial value for the P & I values were idenfied but need fine tuning using actual grinding data

Work Plan for 3 months Grinding tests with single axis oscillation of the workpiece. Vibration No vibration Collecting the data from experimental work for controller fine tuning. Extended experiment for controlled oscillation in grinding.

Thank you Heisum Ewad H.M.Ewad@2011.ljmu.ac.uk