Bosch Ltd

Brief description of the project.

The automotive industry is a vast market where in the recent past there has been a high demand for the diesel operated vehicles for their improved efficiency & reduced emissions. Bosch made Hydraulic Gear Pump has very high demand. Bosch Rexroth plant was unable to fulfill the Customer demand due the bottleneck in Grinding process. To overcome this capacity gap, Bosch Rexroth Plant need to invest 80 million. Since ROI is > 5 years, Bosch management rejected the proposal of new investment. Hence Bosch Rexroth (HejP) approached Bosch Bidadi to improve the productivity by 30%, Bosch Bidadi plant discussed, analyzed & decided to support Bosch Rexroth (HejP) by utilizing the surplus machine, without any additional investment with Bosch Bidadi grinding specialist team. To enhance the process performance, we applied Machine Learning (ML- Supervised Learning Model) techniques to optimize the tool life by 25%. Total revenue generation to Bosch Bidadi Plant is 50 Million INR.

Trigger for the project?


Our main Business vision for the year 2021 is to “Lead in Lean process and excellence in operations”. To keep our self as cost competent and gain profitability, Bosch Bidadi taken challenging task of establishing new Product (Gear Shaft) in < 2 Months during COVID Pandemic.

The following were the project key targets to be fulfilled.

  1. Establishing Gear shaft grinding process in Conventional grinding machine (20 years old) without OEM Support within 2 months during COVID pandemic situation.
    • Cycle time reduction (152 sec/pc 475 sec/pc) by optimizing the multi-stage grinding to single stage grinding.
  1. Gear thickness rejection reduction through Shainin FACTUAL approach.
  2. Tool cost optimization using Machine Learning techniques.

Solution generation, Innovation and Complexity :

Task 1: Establishment of Gear shaft (Multi stage to single stage process)

  • Boisch-HejP approached Bosch Bidadi plant for establishment of Gear shaft due to high demand from Customer & low plant capacity
  • Bosch-BidP had brainstorming with experts internally & finalised to utilise surplus Junker Machines to support Bosch- HejP without any additional investment of Machines within 2months.
  • To achieve concave profile of 5p on Gear shaft ,we need to convert from straight grinding to angular wheel head grinding which calls for 3ONC program modification
  • Due to key elements like axis, tooling's, fixtures & software modifications to fulfil the grinding requirement of Gear shaft, initial proposal rejected by Management for high cost.
  • We took the challenging task of establishing Gear shaft in 20yrs old MAE without OEM support with the stringent time-line <.

Task 2: Reduction of Gear shaft thickness rejection using Shainin - FACTUAL approach:

  • After establishment ,we observed gear shaft thickness variation+/-20p against tolerance of 5 p ,with internal rejection of 5%
  • Applied Shainin — FACTUAL approach to identify the RED X(Root cause) using WOW vs BOB
  • Multi-vari & ISO Plot tools used to analyse the variation & identified the Root cause

Task 3- Adantiya Dynamic Drassmn (ADM to nntimisa ten! Cost-

  • During cost analysis, team found 20INR/pc spend only for Grinding wheels ,Avg. 30Iakh rupees spent/yr
  • Team decided to optimize the tool cost to achieve benchmark cost/pc,frequent manual dressing leads to high tool cost/pc
  • Innovative concept applied using "Machine learning technique" of Adaptive Dynamic Dressing(ADD) in cylindrical grinding
  • After dressing, grinding wheel will be sharp, hence vibration signal during grinding will be less. After grinding —250pcs grinding wheel grains will lose its sharpness which leads to high vibration signal during grinding
  • Taking this contrast data,ML model designed in such a way machine will automatically calls for dressing based on grinding wheel topography.

Implementation of solution :

Task 1: Establishment of Gear shaft :

  • To achieve the 5p thickness toleranceam modified & implemented the existing diameter gauge to thickness measurement gauge without investing 8lakh for new thickness gauge
  • Internally designed & manufactured friction centers to drive gear shaft while grinding, this enable us to integrate 3 stages to single stage grinding
  • Internally designed & manufactured grippers, loading trays, master samples & loading programs this enable us for auto loading & unloading of parts
  •  Quality results are validated through Cm, Cmk study with >1.66 result for all critical

Task 2: Reduction of Gear shaft thickness rejection using Shainin- FACTUAL approach:

  • From Regression analysis, we found that T1 value (thickness value of gauge1) has strong positive co-relation with gear thickness value.  Based on this input, supplier wise incoming parts identification introduced.
  • NC program(as per above pic) developed in such a way, based on incoming parts mean thickness value, in process gauge automatically set T1 value & thereby gear thickness value controlled & produced with minimum variation.

Task 3: Adaptive Dynamic Dressing (ADD) to optimise tool Cost: 

  • To capture & analyze vibration signals we used BOSCH designed IoT gateways
  • Data are pre-cleaned & converted to structured data using MiT Lab software
  • Data categorized using 13time domain parameters & 3 frequency domain parameters
  • These parameters are analyzed & ranked using ANOVA(analysis of variance)
  • Top2 ranked parameters, vibration peek amplitude & vibration standard deviation where taken for model building
  • Degradation model keeps on tracking the grinding cycle in-real time & decision for dressing will be taken by machine learning model based on grinding wheel topography.

Results / Impact:

# Mandatory parameters: Before After Unit of Measurement
1 Finished Goods 2,64,289 4,91,775 pump/annum
2 WIP/ Intermediate goods/ products Nil Nil Nil
3 Scrap generation 0.5 0.01 %
4 Quality a)Gearthickness variation > 20p
b)Cm,Cmk <1
c)Customer complaint : 2000
Gear thickness variation < 5 p
Cm,Cmk >1.66
Customer complaint : <10
5 Direct Cost a)nil
b)Tool cost consumption 20
Revenue generation 90 Tool cost consumption 10 Lakh/annum
6 Manpower cost 3 permanent associate (30Iakh/annum) 3 contract associate (7lakh/annum) no's
7 Delivery 85 98 %
8 Safety Manual loading & unloading Auto loading & unloading no's
9 Internal defect cost 5lakh/annum 0.1Iakh/annum Lakh/annum
10 power saving cost 5.76 1.92 Lakh/annum

Resource impact:

# Parameters: Before After Unit of
1 Energy saving 60 20 Kw/hr
2 Cutting oil consumption 1500 800 Litres/annum
3 Vitrified CBN grinding wheel 6 3 Wheels/annum
4 4Grinding muck generation 6 4.5 Kg/day
5 Compresed air Litres/min
6 Patent registration
Adaptive dynamic dressing using Machine learning technique
0 1 No's

Business metrics :

# Parameters: Before After Unit of
1 Sales volume 2,64,289 4,91,775 Pumps/annum
2 Total turnover 457.39 801.87 Million INR
3 Growth over last year —40 86 %
4 Customer satisfaction scores/ ratings 84 97 %

Leave a Reply