Maintenance Optimization Models

1
Maintenance Optimization Models

• ???

2
Introduction

• The concept of optimization
• The importance of building mathematical models of
  maintenance decision problems
• Key maintenance decision areas
• Component replacement, capital-equipment,
  inspection procedures, resource requirements
• Optimization model
• The role of Artificial intelligence

3
What is optimization all about?

• Optimal
• means the most desirable outcome possible under
  restricted circumstances
• Maintenance decision optimization
• Travel routing problem
• Home Chicago
• Destination London, Moscow, Hawaii
• Item airlines, fares, schedules
• Economy, speed, safety, extras
• Optimize in one area, almost always get a less
  desirable result in one or more of the other
  criteria

4
What is optimization all about?

• Accept a tradeoff
• In any optimization situation, including
  maintenance decision optimization, you should
• Think about optimization when making maintenance
  decisions
• Consider what maintenance decision you want to
  optimize
• Explore how you can do this

5
Thinking optimization

• Thinking about optimization
• means considering tradeoffs-the pros and cons.
• To get every order for every customer delivered
  without fail on the day the customer specified,
  100 of the time.
• A profit-optimization strategy
• the best tradeoff between the cost of inventory
  and an acceptable and competitive customer
  satisfaction level.

6
What to optimize

• You can optimize in maintenance for different
  criteria, including cost, availability, safety,
  and profit.
• Lowest-costs
• The cost of the component, Asset, Labor, Lost
  production,
• Customer dissatisfaction from delayed
  deliveries
• Ex) equipment or component wear-out
• Availability
• Getting the right balance between taking
  equipment out of service for preventive
  maintenance and suffering outages due to
  break-downs.

7
What to optimize

• Safety
• If safety is most important, you might optimize
  for the safest possible solution but with an
  acceptable impact on cost.
• Profit
• If you optimize for profit, you would take into
  account not only cost but the effect on revenues
  through greater customer satisfaction (better
  profit) or delayed deliveries (lower profits)

8
How to optimize

• One of the main tools in the scientific approach
  to management decision making is building an
  evaluative model, usually mathematical, to assess
  a variety of alternative decisions.
• When applying quantitative techniques to
  management problems, we frequently use a symbolic
  model.
• The systems relationships are represented by
  symbols and properties described by mathematical
  equations.

9
How to optimize

• A Stores Problem
• A stores controller wants to know how much to
  order each time the stock level of an item
  reaches zero.

Figure 10-1 An inventory problem
10
How to optimize

• The conflict
• the more items ordered at any time, the more
  ordering costs will decrease
• but holding costs increase, since more stock is
  kept on hand.

Figure 10-2 Economic order quantity
11
How to optimize

• The stores controller wants to determine which
  order quantity will minimize the total cost.
• A much more rapid solution is to construct a
  mathematical model of the decision situation.
• D Total annual demand
• Q Order quantity
• Co Ordering cost per order
• Ch Stockholding cost per item per year
• Total cost per year of Ordering cost
  per year
• ordering and holding stock Stockholding
  cost per year

12
How to optimize

• Ordering cost per year Number of orders
  placed per
• year
  Ordering cost per
• 
  order
• Stockholding cost per year Average number of
  items in
• stock
  per year (assuming linear
• 
  decrease of stock)
• 
  Stockholding cost per item
• Total cost per year
• This is a mathematical model of the problem
  relating order quantity Q to total cost C(Q)

13
How to optimize

• The number of items to order to minimize the
  total cost
• The answer is obtained by differentiating the
  equation with respect to Q, the order quantity,
  and equating the answer to zero as follows

14
How to optimize

• The order quantity equalizes the holding and
  ordering costs per year.
• Example Let D 1000 items, C0 5, Ch 0.25
• No consideration
• Quantity discounts
• The possible lead time (place an order and its
  receipt)
• May Not be linear or known for certain

15
How to optimize

• The purpose of the above model is simply to
  illustrate constructing a model and attaining a
  solution for a particular problem.
• Its clear from the above inventory control
  example that we need the right kind of data,
  properly organized.
• CMMS or EAM system store the vast amount of data.
• It makes optimization analyses possible
• Software is available to help you make optimal
  maintenance decision

16
Key maintenance management decision area

• To build strong maintenance optimization, you
  need an appropriate source, or sources, of data.

Optimizing Equipment Maintenance Replacement
Decisions
Component Replacement
Capital Replacement
Inspection Procedures
Resource Requirements
17
(No Transcript)
18
Key maintenance management decision area

• Resource Requirements
• When it comes to maintenance resource
  requirements, you must decide
• what resources there should be
• where they should be located
• who should own them
• how they should be used
• Your challenge is to balance spending on
  maintenance resources such as equipment, spares,
  and staff with an appropriate return for
  investment.

19
Key maintenance management decision area

• Role of queuing theory to establish resource
  requirements
• The branch of mathematics known as queuing
  theory, or waiting-line theory, is valuable in
  situations where bottlenecks can occur.
• Queuing theory(?? ??)
• ???? ??? ??? ???? ?? ?? ???? ??? ? ???? ???? ???
  ???? ???? ???? ??? ????? ???? ??? ???? ??? ? ??
  ??? ?? ??? ?? ? ??

20
Key maintenance management decision area
Figure 10-4 Optimal number of machines in a
workshop
21
Key maintenance management decision area

• Optimizing maintenance schedules
• In deciding maintenance resource requirements,
  you must also consider how to use resources
  efficiently.
• An important consideration is scheduling jobs
  through a workshop.
• Hong Kong Mass Transit Railway Corporation(MTRC)
• Maintenance Cost and Equipment-failure
  consequences
• Smart scheduling reduces the overall maintenance
  budget.
• GA algorithm 25 reduced

22
Key maintenance management decision area

• Optimal use of contractors (Alternative service
  delivery providers)
• The problem of contracting out the maintenance
  task
• The optimal decision is a balance between
  internal resources and contracting out

23
Key maintenance management decision area

• Role of simulation in maintenance optimization
• How large should the maintenance crews be?
• What mix of machines should there be in a
  workshop?
• What rules should be used to schedule work
  through the workshop?
• What skill sets should we have in the maintenance
  teams?
• Some of these questions can be answered by using
  a mathematical model.

24
Key maintenance management decision area

• Establish the optimal maintenance crew size and
  shift pattern in a petro-chemical plant.

25
Role of artificial intelligence in maintenance
optimization expert systems and neural networks

• Collecting data is very easy.
• Good hardware/good software
• Analyzing the data is more difficult, because
  human experts are required to interpret it.
• Two techological solutions are in varying
  evolutionary states
• expert systems
• neural networks

26
Expert system

• ??? ???? ??
• ?? ?? ?? ?? ?? ??? ???? ???? ??? ??? ???? ?? ???
  ?? ????. ??? ?? ??? ??? ??? ??? ???? ??? ???? ??
  ??? ?? ??? ???? ??? ???? ?? ??? ??.
• ??
• ????? database(?? ??)
• ????

27
Expert system
Table 10-1 An example of an expert system
knowledge base
28
Expert system
Rule IF Machine type is (engine diesel) AND
condition 1(Iron high) AND condition 2(Chromium
high) AND THEN Choice 1 diagnostic (PISTON RING
WEAR/DAMAGE) -7/10 Choice 2 recommendation
(Replace Piston Ring within 100 hours)
9/10 AND ELSE Choice 3, etc
Figure 10-7 An oil-analysis expert system rule.
29
Expert system

• Fuzzy expert system
• ??, ? ?? ??? ???? ???? ?? ??? ??? ???? ??.
• ??? ?? ?????, ??? ???? ?? ??? ???.
• ????? ??? ???? ????? ?,??? ??? ?? ??? ??? ????
  ??? ?????? ???? ??, ??? ????? ??? ???? ?? ???????
  ?? ?????? ????? ?? ??? ? ? ?? ???? ???? ?? ??.

30
Neural networks

• Neural networks observe
• and learn on their own system
• during a training period.
• ?? ? ??
• ?? ??? ??(?? ??) ?
• Class
• ?? ??? ??
• MLP classifier is a popular form

Figure 10-8 A neural network