Scribd
Upload
47 views

Operational Consulting Tools

The document discusses three tools for analyzing manufacturing systems: [1] Queueing theory can provide insights into how changes affect WIP and flowtime as utilization increases. Key ideas are that these metrics are non-linear functions of utilization and variability increases WIP and flowtime. [2] Simulation allows studying complex systems, compressing experiment time, and avoids disruption but requires modeling expertise. Monte Carlo simulation uses empirical data distributions. [3] Concept maps and flowcharts show process flows and interactions to understand actual vs believed processes.

Uploaded by

Samantha Hazell
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
47 views

Operational Consulting Tools

The document discusses three tools for analyzing manufacturing systems: [1] Queueing theory can provide insights into how changes affect WIP and flowtime as utilization increases. Key ideas are that these metrics are non-linear functions of utilization and variability increases WIP and flowtime. [2] Simulation allows studying complex systems, compressing experiment time, and avoids disruption but requires modeling expertise. Monte Carlo simulation uses empirical data distributions. [3] Concept maps and flowcharts show process flows and interactions to understand actual vs believed processes.

Uploaded by

Samantha Hazell
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

The Tools

I. Understanding the Impact of Variability: Queueing Theory


II. Simulation
III. Concept Maps, FlowCharts

I. Basics of Queueing Theory

Queueing Theory can provide critical intuition when analyzing manufacturing


planning and control control systems. Just as human beings queue to watch a
movie, parts queue while waiting to be processed at specific machines.

- number in the queueing system ➪ WIP


- arrival rate λ ➪ production rate
- Sµ ➪ the capacity
- time in the system ➪ flowtime.

Some Notation
λ = Average arrival rate of customers (e.g. customers per hour)
µ = Average service rate capacity of 1 server (e.g. customers per hour)
S = Number of servers
ρ = Utilization Factor or Traffic Intensity ρ = λ / (Sµ)

Performance Measures
Nq = Average number in the queue.
Tq = Average time in queue.
Ns = Average number in the system.
Ts = Average time in the system.
Most Important Results of Queueing Theory

1. Arrival Rate = Departure Rate, on average, in the long run


(departures include "served customers" and those who "balk")

2. Saturation occurs when λ ≥ S µ (if there is no balking)

3. Average utilization of servers = Work Performed / System Capacity


Utilization = ρ ≡ λ / (Sµ))

4. Little's Law:

 Average   Average 
 Quantity in = * Time in 
 System  λ  Sysem 
   

This applies to a whole system or any portion of a system.

Or in manufacturing:
Avg. Number Production Avg. Time Spent
in the = Rate * in the System
System (Throughput) (Flowtime)
(WIP)

Little's Law remarkably holds under a wide variety of systems and


can be used as a rule of thumb to help determine what effects
changes will have on your manufacturing system.

5. Causes of waiting lines: Longer lines are caused by:


a. λ ≥ S µ line grows without limit (or until customers begin to balk.)
b. Variability in either service time and/or arrivals (even if λ ≤ S µ)
c. ρ → 1.0 high average utilization.
Single Server Queue: Number in System vs
Arrival Rate (Mean Service Time 10 Minutes)
100
Mean Number in the System 90
80
70
60
50
40
30
20
10
0
0% 20% 40% 60% 80% 100%
Utilization

Single Server Queue: Time in System vs


Arrival Rate (Mean Service Time 10 Minutes)
1000
Average Time Spent in the System (minutes)

900
800
700
600
500
400
300
200
100
0
0% 20% 40% 60% 80% 100%
Utilization

The key ideas here:


• flowtime and wip are non-linear functions of capacity utilization
• higher variability leads to higher wip and longer flowtimes
• your ability to "absorb" variability depends on capacity utilization
II. Simulation

A. Overview of Simulation

- is a descriptive approach to estimate the


characteristics of a system under various
alternatives for design (or decison variables),

- does not prescribe (tell you what you should do).

- does not duplicate reality, only the most relevant


variables are considered.

When is Simulation Used

- for complex problems for which math is too hard


- when experimentaion is too costly or disruptive
- to compress experimentaion time from years to minutes

B. Advantages and Disadvantages of Simulation

Advantages Disadvantages
Facilitates understanding of Does not suggest a solution
complex systems methodology

Applies to problems that defy Does not apply to deterministic


mathematical solution problems

Avoids risk or disruptive Does not necessarily yield an


experimentation with optimal solution
actual systems

Compresses time to reveal Requires expertise to construct


long-range effects sophisticated models

Less costly than real world Uses expensive labor and


experimentation computer time
C. Monte Carlo simulation using empirical data

Monte Carlo simualtions use random observations from a probability


distribution to duplicate the variability pattern in the system under study.

(i) Collect actual (empirical) data on the distribution of events under


study

(ii) Develop a probability distribution and a cumulative probability


distribution.

(iii) Assign an interval or random numbers to each class of the


distribution.

(iv) Using random numbers (RN’s), derive simulated times times for the
events.

(v) Interpret the results

D. Generating Events Randomly

Random Numbers between 0 and 1, all values equally likely:

1. The Uniform[0,1] Probability Distribution

- pseudo random number generators available (i.e. in Excel)

1
E[X] = .5, Var[X] = 12 s[X] = 0.2887

Prob{X≤.1} = ?
Prob{X≤.77321} = ?
2. Generating other probability distributions
Let R be a "random number" between 0 and 1. (R is Uniform[0,1] )
Use transforms on R

For Example:

- Uniform[a,b]: Y = a + (b-a)R

- Exponential R.V. with Mean = m: Y = -m ln(1-R)

- Normally Distributed R.V. with Mean = m, StDev = s:


Y = m + s • (-6 + Sum of 12 RNs)

E. Performing a Simulation

1. Typical Procedure:
- specify a run-in (burn-in) period during which the
simulation is run but data is not collected
Why?
- specify a run-length during which data is collected
(long enough to average out variation)
- specify the number of runs to be performed
(5 to 10 or more ?)
- specify the random number seeds for each run
Note: Experimentaion is normally to to help determine
the run-in period and the run-length depending on
the required confidence bound for the results

A factory simulator will be made available for individual projects:


- custom software
- Excel and Visual Basic
- FactorAim from Pritsker Corp
III. Concept Maps / Flow Charts

A concept map or flowchart is simply a chart with boxes, diamonds and arrows
used to show the time sequence or flow of parts, processes, orders, information,
etc.

- collecting the information and generating the chart provides critical


information on actual processes and procedures vs those believed
to be in place by management

- useful to help understand and evaluate complex processes

- aids in understanding the interaction of various systems and


subsystems

- typically, the first step used to analyze service and


manufacturing operations from the overall perspective

- be careful of effect of observation on process being charted

Many in-house consulting companies have their own standards for using and
generating flow charts / concept maps, and their own support software

You might also like