Confidential crafting innovation together crafting innovation together CEP course on System Modeling for DRDO Scientists during 18-22 January 2016 at Center for Artificial Intelligence and Robotics (CAIR), DRDO, Bengaluru •Modeling Causality with Event Relationship Graphs •Petri Nets for Dynamic Event driven system modeling By Navneet Bhushan Crafitti Consulting Private Limited Talks on 21 st January 2016
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
CEP course on System Modeling
for DRDO Scientists during 18-22
January 2016 at Center for
Artificial Intelligence and Robotics
(CAIR), DRDO, Bengaluru
•Modeling Causality with Event Relationship Graphs
•Petri Nets for Dynamic Event driven system modeling
By Navneet BhushanCrafitti Consulting Private
Limited
Talks on 21st January 2016
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
2
SYSTEM AND FUNCTION
• System - from fundamental principles point of view – set of elements working together to achieve an
objective or perform a function. – set of elements (Energy-matter organized in space-time) working together
(exchanging energy and information) to achieve an objective or perform a function (Create Change - in Matter, Energy and Information in Space-time).
• When the system is achieved by thought and consciousness we make systems artificial – A TECHNICAL SYSTEM
System
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
… But we can make a Model and Simulate the model of
nearly everything,
potentially … as everything has
Information
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Decisions must be made and actions must be taken today, but the results are not clear until tomorrow, at best.
The uncertainty of the situation at hand makes every decision a burden
Handbook of Foreign Policy Analysis
Play ˀ “If only I know …”
ˀ “If only I had known …”
Decision
Dealing with UNCERTAINTY
To minimize uncertainty, … by taking pains to get more information about the environment
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
SituationAssessment Explanation
Forecasting
Options Generation
Making Decision - Choice
Science is not an oracle – but it can help to reduce uncertainty
Situation Assessment
· Data Collection
· Data Cleansing
· Data Collation· Classification· Observation
Explanation
· Causal Analysis
· Cognitive Mapping
· Systems Analysis
Forecasting
· Historical Analogies
· General Analogies
· Prediction· Projection· Forecasting· “What if”
Analysis
Options Generation
· Decision Trees
· Scenario Writing
· Alternatives
· Brainstorming
· Solution
Choice
· Optimization· Decision
Making under uncertainty and partial information
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Situation Assessment – Where do we stand? Simple Indicators and Checklists/ Complex Indicators/ Scaling (R-factor Analysis)/ Typologies (Q-factor Analysis) /Cluster Analysis /Multidimensional Scaling/ Artificial
Neural Networks (ANN)/ Value Stream Mapping / TRIZ – 9 Windows/ TRIZ- Ideal Final Result
Explanation – Why are things as they are?Correlation Analysis/ Regression Analysis/Analysis of Non-Linear Relationships/ Partial
and Multiple Correlation Analysis/ Multiple Regression Analysis/ Path Analysis
Forecast – What will happen? Systematic Expert Judgment/ Decision Matrix/ Analytic Hierarchy Process/ Bayesian
Inference/ Cross-Impact Analysis/ Early warning Indicators/ Extrapolation with Moving Averages/ Trend Analysis/ Time Series Analysis/ Spectral Analysis/Combined Trend and
Time Series Analysis /Trend Impact Analysis
Preparation of Decisions – What are the Options? Game Theory/ Gaming/ Computer Simulation/ Cellular Automata/ Petri Nets/
Econometric Models/Mathematical Modeling / TRIZ
Choice – What to do? Decisional Trees/ Decisional Matrix/ Linear Partial Information (LPI) Analysis/
c r a f t i n g i n n o v a t i o n t o g e t h e r
19
PETRI NETS
• Graph Models of System behaviour ( a bi-partite graph)• Abstract and Formal• Description and Analysis of – Info and Control Flow• Asynchronous and Concurrent activities can be modeled
System State: Holding of a set of conditionsState Change: End of some conditions and Start of some conditionsEvent: Elementary state change (atomic)
Definition: PNs are graph models for system description using notions of conditions and events
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
PETRI NET Model of a marriage !
Man
P1
Woman
P2
P3Pundit/Qazi/Minister/ Judge
t1Husband
P4
Wife
P5
Pre-Conditions
Post-Conditions
Places
transition
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Analysis Methods• CoverabilityTree• Matrix State Equations• Reduction or Decomposition
techniques
System Simulation
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
A PETRI NET BASED SIMULATION APPROACH FOR EVALUATING BENEFITS OF TIME TEMPERATURE INDICATOR AND WIRELESS TECHNOLOGIES IN PERISHABLE GOODS RETAIL MANAGEMENT
Navneet Bhushan and Kishore
17-18 June 2002, Cork, Ireland
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
TALK ORGANISATION
• PERISHABLE GOODS RETAIL MANAGEMENT- CURRENT SCENARIO AND PROBLEMS
• TIME TEMPERATURE INDICATOR AND WIRELESS TECHNOLOGIES – THE PROPOSED SOLUTION
• PETRI NETS (PN) FOR MODELING & SIMULATION
• PN BASED SIMULATION OF PERISHABLES RETAIL MANAGEMENT
• TEST SCENARIO• PN MODEL OF THE SYSTEM• SIMULATION RESULTS• ANALYSIS• CONCLUSIONS AND FURTHER WORK
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
PERISHABLE GOODS RETAIL MANAGEMENT
• Perishable Goods (PG) – Fruits, Meat Products, Medicines, Chemicals, etc. need Sufficient Cold Storage from the Production stage to the Consumption Stage so as to remain fresh
• Presently a Sell by Date label is fixed by the vendor• No way to find out whether the cold chain was maintained on the
way to retail store or not.• Problem lies in limitations of technology to ascertain the freshness
of the products.
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Potential breakage in the cold chain
Apply a Barcode with “Sell By Date”
Maintain the cold chain
Manage inventory based on FIFO or the “Sell-By-Date
Maintain the cold chain
Manage inventory based on FIFO or the “Sell-By-Date”
Higher shrinkSale of unsafe productSub Optimal use of available life
Vendor
Transportation
Warehousing
Distribution
Retail
The Current Scenario
PERISHABLE GOODS RETAIL MANAGEMENT
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
CAN TECHNOLOGY SOLVE THE PROBLEM ? – A Proposal
• Combining Two Key Technologies • TIME TEMPERATURE INDICATORS (TTI) are capable of
measuring the life of temperature sensitive products - An adhesive label consisting of an enzyme and a substrate filled ampoule separated by a breakable seal. The colour of the ampoule changes from green at the start to yellow at the end of product life cycle. An increase in temperature beyond the specific temperature hasten the color change. Change in color if captured can tell the remaining life of the product.
• WIRELESS LOCAL AREA NETWORKING (WLAN) IEEE 802.11b standard, is a wireless networking technology that can integrate mobile devices to the wired infrastructure as well as to each other through wireless links. WLANs are already being deployed in large stores and organizations.
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Vendor
Transportation
Warehousing
Distribution
Retail
Apply a Barcode with “Sell By Date”
Manage inventory based revised “Sell-By-Date”
Update the Sell By date based on remaining life
Inform Vendor & transporter in case of reduced life
Find cause & take action
Find cause & take action
Retail Category Management systems to help decisions on pricing.
Vendor rating systems to monitor vendor performance
c r a f t i n g i n n o v a t i o n t o g e t h e r
MODELLING AND SIMULATION – PETRI NETS• A powerful modelling framework
for information flow.• Major use is for modelling
Concurrent occurrences with constraints, precedence or frequency of occurrences.
• Carl Adam Petri 1962
Directed, weighted, bipartite graph - two kinds of node, places and transitions.
Places represent conditions and transition represent events.
A transition has input and output places representing pre-conditions and post-conditions of events
Formal Definition: Petri Net (PN)
PN = { P,T,F,W,Mo) P = { p1,p2,…,pm } is finite set of places. T = { t1,t2,…,tn } is finite set of transitions. F { P X T } { T X P } is a set of arcs (flow relation). W : F-> {1,2,3,…} is weight function. Mo : P-> {0,1,2,3,…} is the initial marking. Also, P T = P U T ≠
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
A state or marking in a PN is changed according to following transition rule:• A transition t is said to be enabled if each input place p of t marked with at
least w(p,t).• An enabled transition may or may not fire.• A firing enabled transition t removes w(p,t) tokens from each input place p
of t, adds w(p,t) tokens to each output place p of t.
Simulation of dynamic behavior of systems
Timed Transition Petri Nets• Transition have firing time associated with them.• Time may be Deterministic or Stochastic• The stochastic transition may follow an exponential distribution with
parameter
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Petri Net: Food Selling illustration
Customers in Store
Apples on Shelf
Oranges on Shelf
4
2
Occurrence of Purchase
Satisfied Customer
Firing
Customers in Store
Apples on Shelf
Oranges on Shelf
4
2
Occurrence of Purchase
Satisfied Customer
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
THE SIMULATION SCENARIO
• The Demand: In multiple Retail Stores there has been observed an average demand for 200 Boxes of Perishable Goods per day. Further this demand has been observed to follow a Poisson probability distribution, i.e., the inter arrival time between two demands is exponentially distributed with mean 200 boxes per day.
• The Perishable Goods: The Perishable Goods are labeled with Sell by Date at the Vendor Place. This Sell by Date is computed assuming average cold storage conditions on the way from the vendor to the store. The boxes having the least Sell by Date are picked in a FIFO manner.
• The Vendor: The vendor sends on an average 2000 boxes every 10th day. This is assuming that retail stores have a capability of storing 2000 boxes for 10 days after which the goods expire. These boxes are sent in 20 trucks each carrying 100 boxes. These trucks reach the respective stores on an average in 2 days starting from the vendor to the stores. On an average 10% of these trucks per 10 days do not meet the cold storage requirements. 10% actually exceed the cold storage required by Sell by Date. And 80% meet the cold conditions required by Sell by Date. Also, we assume that the goods not meeting the required cold conditions (we call them Category C) perish on an average in 2 days of reaching the retail outlets. The goods meeting the cold storage (Category B) perish on an average in 8 days of reaching the retail outlets. While goods exceeding the cold storage (Category A) perish on an average in 10 days of reaching the retail outlet.
• Present Scenario: There is no way presently to distinguish between the three categories at the retail outlets. Hence the Boxes are randomly picked (average 200 boxes per day) from Category A, Category B or Category C. This leads to possible customer dissatisfaction if they buy Category C and loss by the store if they sell Category A goods much before their actual Sell by Date.
• Proposed Solution: In the Proposed Solution because of TTI labels, it is possible to distinguish between the three categories and schedule their selling based on Least Shelf Life First Out (LSFO) scheduling
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
PN MODEL OF THE SYSTEM
ColdChainMaintained p = 0.8
BoxesWithVendor
Requested
Boxes
200
Transportation
CheckColdChain
ReachingStore
ColdChain ColdChainNotMaintained p= 0.1
VeryWellMaintained p = 0.1
100 ActuallyPerished
ExceedSellByDate
SellByDate
DemandCreated
Demand
200
PerishedPutOn Shelf
CategoryCSoldCategoryBSold
GoodToSell
CategoryASold
TotalPerished
G1
G2
G3
BoxesSent
BoxesNeeded
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
SIMULATION RESULTS AND RELATIVE CHANGE IN PARAMETERS
Parameters Present Case (E)
Proposed Solution (N)
Relative Change (N/E)
Total Boxes Sent by Vendor (X)
69231 71224 1.029
Sold Category B (Y) 48972 61512 1.256
Sold Category C 5999 - -
Sold Category A 6061 - -
Perished (P) 220 87 0.395
Left (L) 7979 9625 1.206
% Category B sold (100 * Y/X)
70.7 86.4 1.222
% Perished (100 * P/X) 0.318 0.122 0.384
Category B Sales (% terms) have increased by 22%, perished goods have decreased by 60%.
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
ANALYSIS OF THE SIMULATION RESULTS
Parameter Present Proposed Change (%)Boxes Sold (S) 61032 (Category A, B
and C)61512 (Category B
only)0.786
Revenue (S * $ 120) 61032 * 120 = 7323840 61512 * 120 = 7381440 0.786Profit (S * $20) (P) 61032 * 20 = 1220640 61512 * 20 = 1230240 0.786Loss Due to selling
Category C (1)5999 * $10 = 59990 0
Loss Due to Selling Category A (2)
6061 * $5 = 30305 0
Loss Due to Perished Boxes (3)
220 * $100 = 22000 87 * $100 = 8700 -60.455
Loss Due to Loss in Profit because of Perished Goods (4)
220* $20 = 4400 87 * $20 = 1740 -60.455
Total Loss (L= 1+2+3+4)
116695 10440 -91.054
Net Profit (P –L) 1103945 1219800 10.495
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
CONCLUSIONS AND FURTHER WORK
• The present perishable goods retail management results in considerable loss for the retailer due to inability of the system to predict the storage conditions on the way from the vendor point to the retailer
• Combining two recent technologies, i.e., Time Temperature Integrators (TTI) and Wireless LAN (WLAN), we propose a solution that will considerably solve the problem.
• To evaluate the benefits of proposed solution vis-à-vis existing situation we developed a simulation algorithm based on a well-established technique called PETRI NETS.
• The results of the simulation analysis clearly shows the benefits of the proposed solution. In the assumed scenario, it has been shown that the Loss of perishable goods can be reduced up to 90% in monetary terms and Net profit for the retailer can grow by 10%.
• The study indicates the power of using SIMULATION in analyses of conceptualized solutions before designing the actual solution.
• Next step is to design and develop the solution as the benefits are clearly quantified using the simulation model described here.
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Case Study: Methodology paper (Petri Net based simulation of Air Defence System)
Performance Evaluation of an Air Defence System Modeled as a Petri Net
System analysis symposium (CASSA , Bangalore, 1997)
AbstractPetri Nets (PN) are one of the powerful models of information flow. The major use
of PNs has been in the modelling of systems of events in which it is possible for some events to occur concurrently but there are constraints on the occurrence, precedence or frequency of these occurrences. This paper presents a novel approach to model an Air Defence System (ADS) as a PN for performance evaluation. The ADS modeled as a PN is simulated to estimate various performance parameters such as throughput, penetration probability and response time of the system. This performance evaluation tool can be successfully adapted to other systems if the corresponding PN models of such systems are available.
c r a f t i n g i n n o v a t i o n t o g e t h e r
The problem – Air Defence System
One of the major factors contributing to Allied victory over Iraq in 1991 Gulf War was the failure of Iraqi Air Defence System. The results of any future war will depend to a large extent upon the performance of Air Defence Systems (ADS) of respective countries. How the ADS of any country will perform under a given threat is a difficult question to answer. For this purpose there is a need to develop a performance evaluation tool for finding out the effectiveness of the ADS. There are four main techniques usually considered for performance evaluation namely, expert judgement, direct measurement, analytical modelling and simulation. Out of these, simulation provides a flexibility and efficiency that is not possible in other techniques. However, describing a system for simulation purposes requires a much involved modelling exercise [3,5].
The system description – Air Defence System Modeled on
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Petri Nets
• Petri Net (PN) is one of the most powerful formal models of information flow. The major use of PN has been the modelling of systems of events in which it is possible for some events to occur concurrently but there are constraints on the occurrence, precedence or frequency of these occurrences. PNs are a graphical and mathematical modelling tool applicable to many systems. As a graphical tool, PNs can be used as a visual communication aid similar to flowcharts, block diagrams and networks. In addition, tokens are used to simulate the dynamic and concurrent activities of systems. As a mathematical tool, it is possible to set up state equations, algebraic equations and other mathematical models governing the behaviour of systems. The concept of PN was originated in the year 1962 when Carl Adam Petri presented his dissertation. Some time later the work of Petri came to the attention of A.W. Holt who applied these concepts to various concurrent processing concepts.
c r a f t i n g i n n o v a t i o n t o g e t h e r
Petri Nets
Formally a PN is defined as a 5 tuple PN = { P, T, F, W, M0} where
P = {p1 ,p2 ,...,pm} is a finite set of places.T = {t1,t2,...,tn} is a finite set of transitions.F {PT} {T P} is a set of arcs (flow relation).W : F {1,2,3,...} is weight function.M0: P {0,1,2,3,...} is the initial marking.
c r a f t i n g i n n o v a t i o n t o g e t h e r
Petri Nets (the dynamic behaviour model of a system)
The behaviour of many systems can be described in terms of system states and their changes. In order to simulate the dynamic behaviour of a system, a state or marking in a PN is changed according to the following transition (firing) rule :• A transition t is said to be enabled if each input place p of t is marked with
atleast w(p,t) tokens, where w(p,t) is the weight of the arc from p to t.• An enabled transition may or may not fire ( depending upon whether or not
the event actually takes place)• A firing of an enabled transition t removes w(p,t) tokens from each input
place p of t, and adds w(t,p) tokens to each output place p of t, where w(t,p) is the weight of the arc from t to p.
• A transition without any input place is called a source transition, and one without any output place is called the sink transition. A timed transition PN is one in which the transitions have a firing time associated with them. This time may be deterministic or stochastic. If the deterministic transition has zero time it is called immediate. The stochastic transition may follow an exponential distribution with parameter .
c r a f t i n g i n n o v a t i o n t o g e t h e r
Simulation Methodology
• Modeling of the hypothetical ADS as a Time Transition Petri Net (TTPN).
• Modeling of each ACC-SFP pair in the ADS as a petri net object. The ACC-SFP pairs are modeled in such a way that they maintain a recursive structure and hence are easily extended to a series of ACC-SFP pairs connected through CAOC.
• The simulator is modeled as an object interacting with the petrinet objects and obtaining information about transitions that are about to fire.
• Event driven simulation of the events that were generated by the petrinet objects is carried out
• A sensitivity analysis of the simulation results towards various input parameter is carried out.
Throughput 0.93 0.93 0.93 0.90 0.897 0.888 0.896 Penetration Probability
0.007 0.007 0.008 0.032 0.038 0.038 0.04
Using Petri Nets methodology it is possible to study a large number of system parameters. The technique described in this paper can be used to carry out performance modelling and system analysis of any system that can be described as a discrete event dynamic system.
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
S-400 Missile system (Triumf) from Russia – can we simulate it and see how effective it will be in Indian scenarios? Compare it with the
existing air defence system?
The S-400 uses three different missiles to cover its entire performance envelope. These are the extremely long range 40N6, long range 48N6 and medium range 9M96 missile. Each one has different capabilities.
Structure
30K6E administration system: manages 8 divisions (battalions)[9][10][11] 55K6E command and control centre based on Ural-532301.91N6E[12] Panoramic radar detection system (range of 600 km) with protection against jamming. Mounted on an MZKT-7930. 300 targets. Decimetric band (S).[13]
6 battalions of 98ZH6E Surface-to-air missile systems consisting of (an independent combat system for autonomous operation):[14] Each battalion can hit no more than 6 goals on their own.[15]+2 another battalions if they are within range 40 km. 92N6(or 2)E Multi-functional radar (range of 400 km). 100 targets.[16]
5P85TE2 launchers and/or 5P85SE2 on the trailer (up to 12 launchers).Surface-to-air missiles allowed by Russian Presidential decree: 48N6E, 48N6E2, 48N6E3, 48N6DM, 9M96E, 9M96E2 and ultra distance 40N6E.[17]
Own the radars system S-400 this is Active electronically scanned array (official government statement)[18]
The Petri Net based Air Defence modeling and simulation methodology can be used
Increasing Innovation Momentum in the Enterprise• A systematic, time-
bound, flexible, initiative to transform the Enterprise into Invention Empowered Enterprise
• Unlike other such initiatives we work closely with each and every mind of the Enterprise based on scientific methods honed through years of practice
• The Enterprise is self-enabled and empowered to adapt to and in fact architect positive change continuously
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r
Crafitti Consulting: Background
63
• INNOVATION RESEARCH AND CONSULTING in business, science and technology contexts
• Started in June 2008 and was incubated at the NSR Centre for Entrepreneurial Learning at IIM Bangalore
• Crafitti’s frameworks provide potent platforms to innovate in crafting strategies, breakthrough products, new services, technological alternatives, patent portfolios, process design and embedding successful change in organizations.
Confidential
cra
ftin
g i
nn
ov
ati
on
to
ge
the
r
c r a f t i n g i n n o v a t i o n t o g e t h e r