Session 12 - Constraint Management

Operations Management-104

Recap of Previous Session:

Understanding Service Operations Management:

A brief look into India’s services sector

Service Design and applications

India’s digital story and applications

Queueing System

Variability Pooling


Session 12

Constraint Management

Debabrata Ghosh

*Images available in the public domain


Continuing with Variability

Process Time (B)Process Time (A)

106 8 10 12 14

(Constant)(Variable)

3

Item Number

Start TimeProcessing

TimeFinish Time

1 0 14 14

2 14 12 26

3 26 10 36

4 36 8 44

5 44 6 50

Item Number


TimeFinish Time

1 14 10 24

2 26 10 36

3 36 10 46

4 46 10 56

5 56 10 66

Source: Operations and Supply Management, Chase, Shankar, Jacobs and Aquilano, Mc-Graw Hill, 12thEd.,2010


Continuing with Variability

Process Time (B) Process Time (A)

106 8 10 12 14

(Constant) (Variable)

Item Number


TimeFinish Time

1 0 10 10

2 10 10 20

3 20 10 30

4 30 10 40

5 40 10 50

Item Number


TimeFinish Time

1 10 6 16

2 20 8 28

3 30 10 40

4 40 12 52

5 52 14 66

When one process takes longer than the average, the time can not be made up

Rather than balancing capacities, the flow of product through the system should be balanced

4


Re-visiting Gotham City!

What to Produce:

Autobot ‘Gamma’ has decided to launch his own door production company. He

decided to focus on three types of doors according to the needs of the Gotham

city customers. The three door types are A,B,C with selling price per unit of Rs 80,45

and 60 respectively. The raw material costs to manufacture these doors are Rs 70, 30

and 40 respectively. And the production rate of these doors in his factory are

estimated to be 4 units/hour, 5 units/hour and 3 units/hour respectively. Which door

type should Autobot Gamma choose to produce and how much of these door types

should his factory produce, assuming there is enough demand for each type of door

in Gotham City?

5

Approach 1: If Autobot Gamma decides to maximise his revenues/sales , he would simply choose to produce door type A.

Approach 2: If Autobot Gamma decides to maximise his margins, he would choose to produce door type C.


A new approach

6

Approach 3

If Autobot Gamma is worried about the utilization of his factory, based on the production capacities available to him, he would focus on the gross margin/hour which are Rs 40/ hour for type A door, Rs 75/hour for type B door and Rs 60/hour for type C door. Thus, Autobot Gamma may decide to choose B for his new production plans.

How much to Produce:

Consider that Autobot Gamma’s production facility has two machines installed and he now plans to produce all three type of doors A,B and C. The profits from each door are estimated to be 40,30 and 35. The unit processing times in these two machines M1 and M2 are 15, 16 and 12 and 14,11,9 minutes respectively. A total of 2400 minutes of production time is available in the week and the weekly demand for the doors have been estimated to be 70,80 and 60 respectively. How much of each door type should Autobot Gamma decide to produce in his production facility?


How much to Produce – the constrained resource utilization approach

7

Solution Approach:Available weekly production time =2400 minutes

Assuming all demand can be fulfilled:Production time required of M1 = 70*15 + 80*16 + 60*12 = 3050 minutesProduction time required of M2 = 70*14 + 80*11 + 60*9 = 2400 minutes

Given the production time availability of 2400 minutes in a week, M1 is a bottleneck resource.

A bottleneck is any resource whose capacity is less than the demand placed upon it.

A B C

Profit 40 30 35

M1 (time) 15 16 12

M2 (time) 14 11 9

Demand 70 80 60


How much to Produce – the constrained resource utilization approach

8

Considering the bottleneck resource utilization

A B C

Profit 40 30 35

M1 (time) 15 16 12

M2 (time) 14 11 9

Demand 70 80 60

Profit/min. of bottleneck utilization

40/15= 2.66 30/16 = 1.875 35/12 = 2.91

Produce C first. Production time on M1= 60*12 = 720 mins. Balance Production time = 2400-720 =1680 mins. Profit from C = 60*35 = Rs 2100

Produce A next. Production time on M1 = 70*15 = 1050 mins. Balance Production time = 1680 – 1050 = 630 mins. Profit from A = 70*40 = Rs 2800

Produce B last. 16x = 630 => x = 630/16 = 39.375 ~ 39 units. Profits from B = 39*30 = Rs. 1170

Available weekly production time =2400 minutes


Linear Programming Approach to Product Mix Problem

9

Max 40 x1 + 30 x2 + 35 x3

s.t.

15 x1 + 16 x2 + 12 x3 <= 240014 x1 + 11 x2 + 9 x3 <= 2400x1 <= 70x2 <= 80x3 <=60(x1, x2, x3) >= 0


Exercise

A company makes a single product whose weekly demand is 100. Identify the constraint when

• The plant works 8 hours shifts (5 days) and takes 20 minutes to make a product.

• The company takes 30 minutes to make each product

• The supplier can provide raw material at the rate of 15/day

• The supplier can give 25 units per day but the organization takes two days to place the order

10


Claim Settlement Exercise

New Life insurance is a service provider to Medicaid and helps process medical claims. For a

small fee, it performs the entire claims processing operation on behalf of Medicaid.

Processing of a claim consists of the following operations:

1. Claims billed by hospitals arrive by mail to the mailroom clerk who sorts them and places

them on a data entry bin.

2. Data entry assistants check claims for completeness and enter into the system only those

claims which are properly formatted or complete.

3. Claims are assigned to a claim process for initial processing.

4. Processed claims are transferred to a claim supervisor for inspection

5. Claims are returned to original claim processors who issues instructions for settlement of

claims.

11


Claim Settlement Exercise

12

Resource Pool (i) Unit Load (minutes per claim) (Ti)

Number of Units in the Resource Pool (Ci)

Effective Capacity of a resource pool (claims per minute) (Ci/Ti)

Mailroom Clerk 1 1 1

Data Entry Assistants 5 8 1.6

Claims Processor 8 12 1.5

Claims Supervisor 2.50 5 2

The effective capacity of the entire system = 1 claims per minute or 60 claims per hour

The mailroom clerk pool is the bottleneck


Goldratt’s Theory of Constraints Guiding Principles

13

Do not focus on balancing capacities. Instead, focus on synchronizing the flow

The marginal value of time spent at a non-bottleneck resource is negligible. Do

not attempt to reduce time at a non-bottleneck resource.

An hour gained at a non-bottleneck resource is a mirage

An hour lost at a bottleneck resource is an hour of throughput loss

The level of utilization of a non-bottleneck resource is controlled by other

constraints within the system

Resources in the system must be utilized, not simply activated.

A constraint is any element that pre-empts the system from achieving the goal of

making more money.

The transfer batch need not, and many times should not be equal to the process

batch.

Source: Operations and Supply Management, Chase,Shankar,Jacobs and Aquilano,McGraw-Hill(SIE), 12th Edition, 2010


Understanding Drum-Buffer-Rope Scheduling

A manufacturing system (A to C) has three machines and an assembly D. Product P has two raw materials , R and S. R gets processed in A and the B, and goes to assembly. S goes to C and then to assembly. A,B,C and D have processing rates of 40, 20, 30 and 35/hour respectively. The demand for product P is estimated at 25units/hour. Schedule the production of P using DBR approach.

A B

C

Assembly

40 20

30

35

R

S

Drum Buffer of 30 mins

Market Demand

AssemblyBuffer of 10 mins

Shipping Buffer of 20 mins

Assume Process Batch Size = 20 units of Product P

30 mins

60 mins

40 mins

34.2 ~ 35 mins

Release time for R = 175 mins

Release time for S = 105 mins

0-20 mins20-55 mins

55-115 mins115-145 mins145-175 mins

55-65 mins65-105 mins

Source: Lecture Notes on Manufacturing Management Systems, G. Srinivasan, IIT Madras

14


Understanding Drum-Buffer-Rope Scheduling

A B

C

Assembly

40 20

30

35

R

S

Drum Buffer of 30 mins

Market Demand

AssemblyBuffer of 10 mins

Shipping Buffer of 20 mins

Assume Process Batch Size = 20 units of Product P

30 mins

60 mins

40 mins

34.2 ~ 35 mins

Release time for R = 175 mins

Release time for S = 105 mins

Identify the Bottleneck

Provide time buffer so that the bottleneck doesn’t starve

Provide assembly buffer so that non-bottleneck resource doesn’t turn into a bottleneck

Bottleneck determines the system throughput

Communicate upstream as to when and how much of material to release

Balance the Flow and not the capacity

15


A note on Batching

While the previous example has not taken set up time of the 3 machines into account,

in real life systems, set-up times play an important role in manufacturing systems

Set up times reduce effective process capacity.

Operations Managers often plan to increase the capacity of a resource by increasing

the batch size. However, batch sizes typically tend to increase WIP or Finished goods

inventory.

Using Little’s Law, holding flow rate constant, higher inventory levels will mean

increased flow times.

How to then choose a batch size for a process flow?

If the set up occurs at the bottleneck step (process is capacity constrained), it is desirable to increase the batch size as this results in a larger process capacity and therefore, a higher flow rate If the set up occurs at a non-bottleneck step, it is desirable to decrease the batch size, as this decreases inventory as well as flow time.

Source: Matching Supply with Demand, Cachon and Terwiesch, McGraw-Hill, 2012

16

Session 12 - Constraint Management

Documents

type b door

door type c

type c door

door types

type of doors

door production company

service operations management

factory produce