Design and Analysis of a Turbine Blade Manufacturing Cell

1
Design and Analysis of a Turbine Blade
Manufacturing Cell
Term Project Presentation

• MEAE-6960H01
• Professor Ernesto Gutierrez-Miravete
• Presenter Ray Surace

2
Project Overview

• Created a hybrid coding scheme for three turbine
  blade part numbers (P/N) to be produced in a
  group technology environment
• Reviewed the performance of an initial turbine
  blade cell design with 12 workstations
• Developed an improved cell design with 10
  workstations
• Created a facility layout using the Column-Sum
  Insertion Heuristic
• Used a modified Approximate Three-Stage Markov
  Chain Model to determine the optimum size of
  buffers placed before and after an airfoil
  overlay coater
• Performed a Mean Value Analysis to validate the
  final design of a turbine blade manufacturing cell

3
Part Coding

• Three (3) P/Ns with common features were grouped
  to form a composite part family
• An alphanumeric hybrid coding scheme was derived
  from the machining sequence
• The code makes use of the chain property each
  character place in the code has a specific
  meaning Example P/N
  100b1
• Code 1 b 1 y 3
• first digit
  part stage cooling
  hole
• of engine model i.e.
  1STG1 code
• part type coating?
• bblade
  yyes, nno
• Cooling hole code 0no cooling holes, 1laser
  cooling holes, 2EDM cooling holes, 3laser and
  EDM cooling holes

4
Performance of 12 Workstation Cell Design

• Initial cell design incorporated 12 workstations
  into a U-shaped cell layout to complete
  machining and finishing operations on P/N 100b1,
  200b1, and 100b2 turbine blade castings.
• 1 2 3 4
  5 6
• isle 12 11 10 9
  8 7

5
Performance of 12 Workstation Cell Design

• Customer demand rates for all 3 P/Ns dictate that
  9.12 parts must be produced per hour
• Thus, cycle time C, must be 0.11 hrs. for each
  workstation
• Processing time of the precipitation heat treat
  furnace is 24 hours the retort can hold 150
  blades. Thus, C 0.16. This is
  unacceptable.
• To meet customer demand rates the furnace retort
  must have a capacity of
• Checking utilization we find that the heat treat
  furnace is a bottleneck operation
• Thus, Um, furnace 1.46
• If a furnace with a capacity of 219 blades is
  purchased, WIP would increase by 69 pcs
• 219 blades - 150 blades 69 blades
• By removing the heat treat furnace from the cell,
  the overall WIP (WIPPxT) of the cell can be
  reduced by 150 pcs.
• New cell design to include 10 workstations
  precipitation heat treat and shot peen machines
  moved to a separate department

6
Layout of 10 Workstation Cell Design

• Final turbine blade manufacturing cell design
  incorporates 10 workstations into a U-shaped
  cell layout
• coater input buffer
• 5 4 3 2 1
• 6
• coater output 7 8 9
  10
• buffer isle

7
Turbine Airfoil Manufacturing Facility Layout

• Column Sum-Insertion heuristic used to create a
  block plan of a turbine airfoil manufacturing
  facility with the following departments
• Receiving
• Vendor Inspection (incoming casting inspection)
• Blade Cell
• Vane Cell
• Heat Treat / Sot Peen Department
• Shipping
• Receiving
• A Receiving
• Vendor Inspect
  I
• A I
  Shipping Blade
  Cell Vendor Insp.
• Blade Cell
  E X
• U U E
• Vane Cell
  A O
• A O
• Heat Treat/Peen
  O Heat Treat /
  Peen Vane Cell
• A
• Shipping

8
Analysis of Coater Buffer Capacity

• Buffers are required before and after the coater
  in order to maintain a cycle time of 0.11 hrs.
• An Approximate Three-Stage Markov Chain Model was
  modified to determine the optimum buffer size
• The following modeling assumptions were made
• Cell modeled as a paced transfer line with M10
  stages
• Coater capacity of 19 blades
• Assumed the average mean time to failure, a-1 of
  workstations 1-5 and 7-10 approaches 0.
  Therefore a1-5,7-10 1e-6
• Assumed acoater1. Once the coater starts a
  cycle, any incoming parts go into the incoming
  buffers
• The avg. mean repair time (coater cycle time)
  for the coater b-118.18cycles, or b0.05
• The results of the Three-Stage model analysis are
  as follows
• The effectiveness of the cell without buffers,
  E00.53282
• The maximum cell effectiveness is E21.53284
• Therefore, the optimum buffer size before and
  after coating is Z21

9
Performance of 10 Workstation Cell Design

• A Mean Value Analysis (MVA) was performed to
  validate the 10 workstation blade cell design
• The cell was modeled as a closed network with
  single servers
• It was assumed that each P/N (p) may visit each
  workstation (j) in that parts processing
  sequence once
• A visit count (Vjp) table was constructed
• To initialize the algorithm, a queue length
  (Ljp) of 1 was assumed in front of each
  workstation
• The algorithm was calculated using the following
  formulas
• (eqn. 11.24 AS) ,
  (eqn. 11.26 AS) ,
  (eqn. 11.27 AS)
• After three (3) iterations the algorithm
  converged
• The total production rate of the cell,
  Xtotal9.92 parts per hour. This exceeds the
  demand, 9.12 parts per hour by 8. Thus, the 10
  workstation cell design is acceptable.