optimization of painting parameters to obtain the desired
TRANSCRIPT
optimization of Painting Parameters to obtain the Desired Gloss value on Metro Car Interior Panels
BEMl ltd.
Sumanta Kumar Saha, G.R. Keshavan, niranjan Kumar Singh, K.S. Sekhar, y. veerakumarRail & Metro Group, BEML Ltd, Bangalore.
Keywords: Design of Experiments, Six Sigma, optimization, Gloss value, Control Chart, Process Capability, normality, Process Mapping, Cause & Effect
AbstrAct
BEML is manufacturing metro coaches for Delhi, Bangalore, Jaipur and Kolkata Metro Rail Corporations. The Interior of these coaches are fitted with GFRP based panels. Earlier these panels used to come in completely painted, in ready to fit condition from M/s. ROTEM, Korea at a high cost. As a part of import substitution and “MAKE IN INDIA” policy, large panel painting facility was established indigenously.
With the painting process given by the collaborator, the required gloss value which is a critical indicator of aesthetics could not be achieved. Hence, the rejection by the customer was high. To optimize painting process parameters to get the desired results, six sigma methodology was adopted.
Introduction
metro cars have revolutionized the concept of mass transit system in metro cities. BemL forayed into manufacturing, testing, supply & commissioning of metro cars since 2001 and has emerged as a strong player in this business. BemL is a leading manufacturer of metro coaches for Delhi, Bangalore, jaipur and Kolkata metro rail corporations. the interior of these coaches are fitted with GFrP based panels as shown in Fig.1 & 2.
Fig.1: Showing GFRP based Interior panels
60 BEML Ltd.
Fig.2: Showing Interior view of a Metro Car
the metro cars are unique and have superior features like – fully fabricated out of austenitic stainless steels, light weight, aesthetically pleasing FrP interiors, regenerating braking system which is environment friendly, automated signaling etc. During the initial orders of Delhi metro, the entire FrP interior panel was imported from South Korea. However, BemL management took a decision to indigenize the FrP interior panels.
the following diagram in fig.3 indicates the various components and stages in manufacturing of FrP interior panel:
Process Mapping
1. Raw material 1a. Check for damage1b. Clean with acetone
2. Sanding – 120 grade
3. Putty application – Drying naturally for 4 hrs
4. Putty sanding after drying
5. Primer preparation epoxy paint #Hardner (3:1 ratio) and thinnerto achieve viscosity
of 24 to 28 sec
7. Check compressed air pressure (5-6 bars)
6. Preparation of gun6a. Select nozzle (1.5 mm)6b. Check for cleanliness
of nozzle
8.1st coat primer spray in booth
11. Putty application & drying
9. Drying in the oven
12. Sanding with 180 gritpaper & cleaning
15. Sanding with 220 grit paper
13.2nd coat primer & drying in oven
14. Putty application & drying
16. Finish paint preparation (4:1 & viscosity of 20 to 24)
16a. Stirring & Filtering
17. For application of 2 coats finish paint spray.Repeat process no. 6,7,9
Move toassy.
Repeat from 2
18.Inspection
10. Inspection
NO
NO
YES
YES
Fig.3: Indicating various components and stages in manufacturing of FRP interior panel
Problem Definition
Gloss value is a critical indicator of interior aesthetic appearance of the metro coaches. Due to variation observed in Gloss value of interior painted panels, approx. 50% panels were being rejected.
Gloss refers to the shine or polish on a smooth surface. In paint technology, Glossy and flat (or matte) are typical extreme levels of glossiness of a finish.
Glossy paints are shiny and reflect most light in the specular (mirror-like) direction, while on flat paints most of the light diffuses in a range of angles.
Between those extremes, there are a number of intermediate gloss levels. their common names, from the most dull to the most shiny, include: matte,
Optimization of Painting Parameters in Metro Car Interior Panels 61
eggshell, satin, silk, semi-gloss and high gloss. these terms are not standardized and not all manufacturers use all these terms.
Goal Statement
to keep the gloss value in between 55 to 75 and improve sigma rating.
Methodology
the methodology used to solve the above problems is Six sigma (DmaIc) with Doe. the details are given in the following pages.
Data Collection Plan
Action: To get the desired value of gloss value while interior panel painting
Data Operational Definition and ProceduresWhat Measure
Type/ Data Type
How Measured
Related Conditions to Record
Sampling Notes
How/ Where Recorded
Gloss Value in Panel Painting
Continuous Using Gloss Meter
Painted interior panels in painting booth in Jan '16
100% Using gloss meter, gloss value measured and recorded
How will you ensure consistency? Data collected by directly measuring samples after establishing Gauge R&R.
What is your plan for starting data collection? Collect data by directly measuring gloss value on painted panels.
Gloss is measured using a Gloss meter which directs a light at a specific angle to the test surface and simultaneously measures the amount of reflection.
Fig. 4: showing how to check gloss value using gloss meter
Gauge r & r study was conducted to ensure the reliable data collection.
62 BEML Ltd.
Fig. 5: Showing Gauge R & R study
as the error is 1.06, Gauge r&r is acceptable to proceed with data collection. Further, the data was analyzed to identify whether it fits any known statistical distribution. the following statistical distributions in fig.6 were tested – Normal, Lognormal, exponential, Weibull distributions.
Fig. 6: Showing probability plot from gloss value
as the p-value is more than 0.05, gloss value follows 2-parameter exponential distribution. the initial sigma rating was calculated and found to be 1.4.
Optimization of Painting Parameters in Metro Car Interior Panels 63
Analysis of the Problem
Various brain storming sessions were conducted to find out the probable root causes using root cause analysis in fig.7 under the heading of machine, method, personnel, measurement and material that lead to variation in gloss value in GFrP based interior panels.
Fig.7: showing cause & effect analysis
all the probable root causes have been validated using different methodologies like Gemba investigation, data analysis using control charts, and verification of various records like painter qualification, calibration records etc to find out the potential root causes.
after detailed investigation, it was found that the potential root causes confirms to the desired values. to further improve the process and to understand the effect of certain parameters on gloss value, a Design of experiments was planned and conducted.
Experimental Factors
Hardener mixing ratio, Heating temperature, Heating time, Spraying Viscosity & Surface Preparation
Control Factors
Paint Source, Nozzle Size, Spraying Pressure, Painter
Cause and Effect Diagram
METHODMACHINE
PERSONNEL MEASUREMENTMATERIAL
Oven not
Calibrated NozzleSize Incorrect
Surface Preparation
Spraying Viscosity
HardenerMixing Ratio
Booth Condition(Filter, Light,
Ventilation etc.
Skill of the Painter
WrongInstrument
Moisture inComp. Air
SprayingPressure
HeatingTemperature
HeatingTime
Stirring ofPaint Mixture
Contamination of Paint
Improper PaintStorage
PaintSupplier
Instrument notCalibrated
Moisture inPaint
Wrong Methodof
MeasurementNo Work
Instruction toPainter
PainterFatigue
Operator toOperator Difference
VARIATION IN GLOSS
64 BEML Ltd.
noise Factors
compressor Supply Line, ambient conditions (temperature, Humidity etc.), operator Fatigue etc.
No. of Factors : 5
Levels : 2
It was planned to conduct a full factorial experiment. Factors and levels were selected as per table 2.
Table 2: Showing the Factors and Levels of DOE
S.No. Factors Low High1 Hardner Mixing Ratio 4:1 8:12 Heating Temperature (in º C) 60 803 Heating Time (in Minutes) 30 604 Spraying Viscosity (in Seconds) 16 255 Surface Preparation (in Grade) 200 500
Type of experiment: Full Factorial, Replicates: 2Randomised to reduce the effect of any lurking variablesCentre Points: Nil, No. of Blocks: 1No. of experimental runs: 64
total 64 experiments were conducted and data were collected for hardener mixing ratio, surface preparation, spraying viscosity, heating time and heating temperature. adequacy of the model was conducted using residual plot and found that residuals follows normal distribution curve and does not show any pattern with the time and fitted value.
time series plot does not show any trend. It can be concluded from the residual plots and time series plot that the model is adequate.
Optimization of Painting Parameters in Metro Car Interior Panels 65
Fig.8: Showing large effects of variables.
Large effects were determined using Pareto chart and normal plot. Significant factors affecting found to be Hardener mixing ratio, Heating temperature, spraying viscosity and Drying time. Subsequently main effect plot was plotted and found that there is a significant change in gloss value when hardener mixing ratio is changed from 4:1 to 8:1. Surface preparation has no effect on gloss value. Interaction plot also plotted for gloss using all five variables and found no interaction.
Finally cube plot is made as the summary of the analysis in Fig. 9.
Fig.9: Showing various combinations of variables to get the desired gloss value.
From Fig.9, shows the levels of painting process parameters to get the optimum gloss value i.e.
66 BEML Ltd.
i) Hardener mixing ratio= 8:1
ii) Heating temp. = 80 degree c
iii) Drying time = 60 minutes
iv) Spraying Viscosity = 16 Seconds
after this, 2t test were conducted and found that new method is better than old method as the p value is less than 0.05.
From the Doe study, optimal levels of process parameters were arrived. a risk analysis was carried out to ensure that the new parameter levels do not affect the requirements of other ctQs. adhesion, paint flow, air bubble etc. checked for 05 samples and found acceptable. No other risks are expected. I & mr process control chart was drawn to find out whether the process is under statistical control or not at new levels of variable. as all the points were within the control limit and no abnormal variation were observed, the new process was found to be in statistical control. the chart below in fig.10 shows the Sigma levels before and after Doe study.
Sigma Rating Before and After
Process Capability Report for Old Gloss ValueCalculations Based on Exponential Distribution Model
Before – 1.4 After – 5.28
55 60 65 70 75 80 85
57 60 63 66 69 72 75
Process Capability Report for New Gloss ValueLB Target UB LSL USL
LB 55Target 65UB 75Sample Mean 77.0625Sample N 16Scale 3.37333Threshold 73.6892
Observed PerformancePPM<LB 0.00PPM>UB 562500.00PPM Total 562500.00
Process DataLSL 55Target *USL 75Sample Mean 64.248Sample N 16StDev (Overall) 2.43061StDev (Within) 2.55513
Process Data
Observed Expected Overall Expected WithinPPM < LSL 0.00 70.96 147.65PPM > USL 0.00 4.85 12.88PPM Total 0.00 75.81 160.53
Performance
OverallWithinPp *
PPL *PPU *Ppk *
Overall Capability
PPM < LB *PPM > UB *PPM Total *
Exp. Overall Performance
Pp 1.37PPL 1.27PPU 1.47Ppk 1.27Cpm *
Cp 1.30CPL 1.21CPU 1.40Cpk 1.21
Overall Capability
Potential (Within) Capability
Fig.10: Showing sigma rating before and after Design of Experiments
Conclusions
above study has resulted in improving the quality of GFrP based interior panels, Higher productivity, customer satisfaction and last but not the least, substantial financial savings by way of imports substitution.
Acknowledgements
the authors express their thanks to the management of BemL Ltd. for the support & guidance in successful completion of the above project.
s
Optimization of Painting Parameters in Metro Car Interior Panels 67