The GRAI method Part 2: detailed modelling and methodological issues
B. Vallespir, G. Doumeingts
The GRAI method, Part 2: detailed modelling and methodological issues
1
Content
• • • • •
1. 2. 3. 4. 5.
Introduction The GRAI nets The structured approach The rules of inconsistencies The GRAI methodology
The GRAI method, Part 2: detailed modelling and methodological issues
2
First part
INTRODUCTION
The GRAI method, Part 2: detailed modelling and methodological issues
3
Definition of the GRAI method
• The GRAI method owns to the enterprise modelling domain. The purpose is to design or reengineer production systems (manufacturing or service). • The GRAI method focuses on the decisional aspect (control system). • From a general point of view, the GRAI method applies to performance improvement.
The GRAI method, Part 2: detailed modelling and methodological issues
4
Composition of the GRAI method The GRAI Method: • is built up starting from a reference model, the GRAI model, which is a consistent set of concepts that model any production system in a generic way and a priori, • is based on graphical modelling languages which instantiate the concepts of the GRAI model to build the specific model of the studied case, • follows a structured and participative approach within which actors and steps are defined, allowing effectiveness and time saving. The GRAI method, Part 2: detailed modelling and methodological issues
5
Composition of the GRAI method
Note: The GRAI reference model and the GRAI grid are not presented in this course and can be found in: The GRAI Method Part 1: global modelling
The GRAI method, Part 2: detailed modelling and methodological issues
6
Application domains of the GRAI method • Production systems engineering, • Choice and implementation of software packages for management: ERP (Enterprise Resources Planning), SCM (Supply Chain Management), CRM (Customer Relationship Management) or other computerized solutions (decisional...); • Choice and implementation of performance indicators systems; • Development and implementation of industrial strategies; • Support to quality approaches; • Knowledge Management.
The GRAI method, Part 2: detailed modelling and methodological issues
7
Application domains of the GRAI method
Remark: The GRAI method takes part in these applications without being sufficient enough in general ⇒ necessity to increase the modelling domain (see GRAI methodology, 5th part)
The GRAI method, Part 2: detailed modelling and methodological issues
8
Second part
THE GRAI NETS
The GRAI method, Part 2: detailed modelling and methodological issues
9
INTRODUCTION
The GRAI method, Part 2: detailed modelling and methodological issues
10
Definition
The GRAI Nets represent the running of the whole or a part of a decision centre according to the GRAI modelling concepts (model of a decision centre activities).
The GRAI method, Part 2: detailed modelling and methodological issues
11
Situation of GRAI nets in modelling F1
F2
F3
...
Fn
H/P H/P H/P
Principle: There is only one network by decision centre
1
Urgent order
MRP
Workload over 3 months
To centre the plan over 3 months
2
PL/30
MR/10
Workload over 3 months completed
To complete workload
Objective: Respect of delivery dates
Progress report on the production DV
Practically: Sometimes several (size, version, etc.)
PL/30 TO PLAN
Rules Priorities Requirement origin MT Workload
PL/30
H = 3 months P = 1 day
To plan workload + ST schedule #
• Delivered quantity • Int resour.
3 To adjust load (smoothing)
Workload for MT
PL/30
Methodological information
PL/20 PL/30
Man. order unreach able
Load distributed per period and station
The GRAI method, Part 2: detailed modelling and methodological issues
PL/10
12
Graphical representation of an activity
Support
Trigger
#
Activity
Result
Support
The GRAI method, Part 2: detailed modelling and methodological issues
13
MODELLING COMPONENTS
The GRAI method, Part 2: detailed modelling and methodological issues
14
Activity and entity Activity. Part of the behaviour of a decisionmaking centre. An activity is dynamic and provides one or more entities. The activities are named and numbered.
Name
Name
Entity. Physical or abstract object that belongs to the control system. The entities are necessary to the course of the activities or are produced by those ones. The entities are named.
The GRAI method, Part 2: detailed modelling and methodological issues
15
Activities type Two types of activities exist Decision activity Execution activity
Trigger
# Support Support
Trigger
# To execute
Support
Result
T o d e c i d e
Support
Result
The GRAI method, Part 2: detailed modelling and methodological issues
16
Activities type Execution. Deterministic activity. Activity giving the same value to the result for the same values of the convergent entities (trigger and supports). Example: activity managed completely by a rule, procedure, programme, etc. Decision. Activity being able to give several values to the result for the same values of the convergent entities. Example: choice of a solution in uncertain context The GRAI method, Part 2: detailed modelling and methodological issues
17
Nature of entity There are seven natures of entities: • Objective. Level of performance attended after the realization of the activity. • Decision variables. Element on which one can play during the activity implementation. • Criterion. Help to the choice of the actions on the decisions variables. • Rules. Specification of the behaviour of an activity or a part of this one.
The GRAI method, Part 2: detailed modelling and methodological issues
18
Nature of entity There are seven natures of entities (cont’): • Performance indicator. Report on a performance. • Information. Entity of informational nature unspecified. • Resource. Concrete means, technical or human, necessary to the implementation of the activity. The nature of the entity is mentioned (or be an information by default) Rule
The GRAI method, Part 2: detailed modelling and methodological issues
19
Role of entity Three roles are possible for an entity Support. Entity required for the progress of an activity Result. Entity produced by an activity Trigger. Entity required for the progress of an activity and of which the disposal triggers the activity.
The GRAI method, Part 2: detailed modelling and methodological issues
20
Additional information Trigger • The trigger of an activity can also consist of an exogenous logical condition. In this case, this one is indicated as an entity trigger. • The trigger can also consist in the period of the level where the decision centre is located. In this case, no trigger is mentioned. Support • Objectives, decision variables and criteria are only and mandatory represented for decisional activities. The GRAI method, Part 2: detailed modelling and methodological issues
21
Reference operators Reference operators are used when it is necessary to indicate the origin or the destination of an entity when this origin or this destination is external with the diagram. Situations for which these operators are necessary: • when an entity circulates between the studied system and its environment, • when an entity circulates from one decision centre to another one, • when the net is too large to enter on a page! The GRAI method, Part 2: detailed modelling and methodological issues
22
Reference operators
Origin : Destination :
R/A
Entity
Entity
R/A
• R: number of the net, • A: number of the origin or destination activity • R/A: name of the system or service out of the study domain The GRAI method, Part 2: detailed modelling and methodological issues
23
LOGICAL OPERATORS
The GRAI method, Part 2: detailed modelling and methodological issues
24
Generalities • It appears combinations or decompositions in the GRAI nets which result in divergences and convergences into AND and OR, from entities to activities and reciprocally. These situations are represented by logical operators. • The representation of these operators follows the general rule: • AND are represented by double features (| |) • OR are represented by simple features (|)
The GRAI method, Part 2: detailed modelling and methodological issues
25
Generalities Example OR Convergence
AND Divergence
The GRAI method, Part 2: detailed modelling and methodological issues
26
Particular case Link: Entity → Activity / AND / convergent
= Because supports are not optional
The GRAI method, Part 2: detailed modelling and methodological issues
27
Prohibited structure Link: Entity → Activity / OR / divergent Entity Entity
⇒
C h o i c e X X
The outcome of the entity to one or the other of the activities supposes the idea of a choice: this implies to clarify this choice The GRAI method, Part 2: detailed modelling and methodological issues
28
Prohibited structure Link: Activity → Entity / AND / convergent
An activity must have its own result Link: Activity → Entity / OR / convergent
The GRAI method, Part 2: detailed modelling and methodological issues
29
EXAMPLE OF NETS
The GRAI method, Part 2: detailed modelling and methodological issues
30
Example
1
MRP
Workload over 3 months
To centre the plan over 3 months
Urgent orders
2
PL/30
MR/10
Workload over 3 months completed
to complete workload
Objective: Respect of delivery dates
Progress report on the production DV • Delivered quantity • Int resour.
PL/30 H = 3 months P = 1 day
TO PLAN
#
To plan workload + ST schedule
3 To adjust load (smoothing)
Workload for MT
PL/30
Rules Priorities Requirement origin MT Workload
PL/30
PL/20 PL/30
Methodological information
Man. order unreachable
Load distributed per period and station
The GRAI method, Part 2: detailed modelling and methodological issues
PL/10
31
Link with the corresponding grid Example of a control grid
50
2 years 1 month
40
8 months 1 month
Forecasts of sales per families
To manage products
To plan
To manage resources
To make Long Term plan
• To define engagement strategy • To define structural S/C
To man. purchase To man. procurem. • To look for suppliers • To negotiate markets
• To define proc. strategy • To define critical proc.
Consolidated orders
Internal information
MSP MRP
To make proc. plan
To plan workload + MT schedule
To define conjectural S/C
To plan workload + ST schedule
To assign the personnel
Orders book
30
6 month 1 week
To send orders to suppliers
20
3 months 1 day
To recall suppliers
10
1 day RT
To define conjectural S/C
To define proc. parameters
Urgent orders To dispatch To record orders
Inventories level
Example
To record I/O raw materials, materials and FP
Workload for MT
PL/30
Title
Date
76
1
MRP
Workload over 3 months
To centre the plan over 3 months
Urgent order
2
Objective: Respect of delivery dates
PL/30
H = 3 months P = 1 day
TO PLAN
MR/10
#
PL/20
Methodological information
Progress report on the production DV • Delivered quantity • Int resour.
PL/30 TO PLAN
To plan workload + ST schedule
Workload over 3 months completed
to complete workload
#
To plan workload + ST schedule
H = 3 months P = 1 day
Rules Priorities Requirement origin MT Workload
PL/30
PL/20 PL/30
Methodological information
Date
3 To adjust load (smoothing)
External information
Man. order unreachable
Load distributed per period and station
Title
The GRAI method, Part 2: detailed modelling and methodological issues
PL/10
31
32
Example: links with the corresponding grid
1
MRP
Workload over 3 months
To centre the plan over 3 months
Urgent order
2
PL/30
PL/40
MR/10
MRP
Workload over 3 months completed
to complete workload
Objective: Respect of delivery dates
Progress report on the production DV
To plan workload + MT schedule PL/30
To recall suppliers GP/20
Inventories level
PL/30
Rules Priorities Requirement origin MT Workload
PL/30
To assign the personnel
To plan workload + ST schedule PL/20 PL/10 To dispatch
• Delivered quantity • Int resour.
3 To adjust load (smoothing)
MT Workload
PL/30
GR/20
PL/30
Man. order unreachable
Load distributed per period and station
The GRAI method, Part 2: detailed modelling and methodological issues
PL/10
33
Example: links with the corresponding grid Consistence GR/10
To recall suppliers GP/20
Absent of the grid: normal (additional detail at the level of the nets)
Progress report of the production
To plan workload + ST schedule PL/20
To assign the personnel GR/20
Absent relations in the net: abnormal
The GRAI method, Part 2: detailed modelling and methodological issues
34
Example: links with the corresponding grid Analysis of the content of the decision framework PL/30 → PL/20 PL/30
In the grid: To plan workload + MT schedule PL/30
To plan workload + ST schedule
Objective Respect of the delivery date DV
Transmission of the value of the objective
PL/30
• delivered quantity • Int resour.
Transmission of the value of the constraints
PL/30
MT Workload
Order transmission
PL/30
Man. order unreachable
PL/20
Information follow up
The GRAI method, Part 2: detailed modelling and methodological issues
35
Third part
THE STRUCTURED APPROACH
The GRAI method, Part 2: detailed modelling and methodological issues
36
General principles
Models
OBJECTIVES
Analysis
Design
Implementation Real world
Modelling
?
Existing system (As-is)
Target system (To-be)
The GRAI method, Part 2: detailed modelling and methodological issues
37
The actors BOARD GROUP
is & Analys on i validat
Define the objectives and the domain of the study Evaluate and validate the results
Synthesis group (Main deciders)
GRAI Specialists (sensors)
Perform the study als Propos Definition Valida Proposals Analysis & tion validation Support the study Inform ation Va Working lid ati groups on Inf Interviewees orm Look for ati on solutions Provide information
The GRAI method, Part 2: detailed modelling and methodological issues
38
The actors Example of a group composition (study of a SME)
• • • • • •
Board group person in charge of the SME responsible for the division of the group to which the SME is attached Synthesis group person in charge of the SME responsible for procurement / subcontracting responsible for manufacturing …/… The GRAI method, Part 2: detailed modelling and methodological issues
39
The actors Example of a group composition • • • •
responsible for quality responsible for estimation / preparation responsible for scheduling responsible for commercial Specialists • a specialist from a service company • a specialist assistance provided by the group
The GRAI method, Part 2: detailed modelling and methodological issues
40
General progress Initialisation Initialisation Modelling Modelling and and existing existinganalysis analysis
Context Context and andobjectives objectives of ofthe thefuture futuresystem system
Design Designof ofthe the future futuresystem system Actions Actions plan plan The GRAI method, Part 2: detailed modelling and methodological issues
41
Course of the phase « initialisation » Presentation Presentationof ofthe the study study phases phases Synthesis group Board Group Groups Groups definition definition
The GRAI method, Part 2: detailed modelling and methodological issues
42
Course of the phase «modelling and existing analysis» Global Globalmodelling modelling
Detailed Detailedmodelling modelling
Analysis Analysis assessment assessment
• Grid • Interviews planning
• Interviews • Setting form (realisation of the nets) • Review of the grid
• Determination of the inconsistencies • Analysis report
Synthesis group
Interviewees
Synthesis group
Synthesis group Board group
The GRAI method, Part 2: detailed modelling and methodological issues
43
Course of the phase « context and future system objectives » Definition Definitionof ofthe the external externalconstraints constraints
Synthesis group
Definition Definitionof ofthe the internal internalconstraints constraints
Synthesis group
Description Description of ofobjectives objectives of ofthe thefuture futuresystem system
Synthesis group Board group
The GRAI method, Part 2: detailed modelling and methodological issues
44
Course of the phase « future system design » Design Designinitialisation initialisation
Global Globaldesign design
Detailed Detaileddesign design
• Inconsistencies solving • Proposals for orientations • Choice of one orientation
Synthesis group Board group
• Grid • Global nets
Synthesis group
• Detailed nets
Working groups
Working groups
Synthesis group
Design Designsynthesis synthesis
• Design report Board group
The GRAI method, Part 2: detailed modelling and methodological issues
45
Some precisions
• The phases do not proceed a such sequential way (overlapping, looping); • It is necessary to regularly gather the group of synthesis (to remain "in catch" with the study) without overloading its members (1/2 day every 2 or 3 weeks): that involves intermediate versions of the results.
The GRAI method, Part 2: detailed modelling and methodological issues
46
Example of planning
The actors
Modelling / Diagnosis 3/4 months maximum
Design 3/4 months maximum
Board Group Synthesis Group Interviews Working Groups
Double arrow = Framework
The GRAI method, Part 2: detailed modelling and methodological issues
47
Fourth part
THE INCONSISTENCIES RULES
The GRAI method, Part 2: detailed modelling and methodological issues
48
Generalities about inconsistencies rules • The most important rules come from the GRAI model (in this way, they are a part of the model); • The inconsistencies rules correspond to the ideal vision of a control system; • The goal is not to obtain such an ideal system, the studied system will not be requested to match all the rules:
⇒ Set of inconsistencies rules = Guide to think about the consistence of the studied system. The GRAI method, Part 2: detailed modelling and methodological issues
49
«A» rule Horizons must be sliding. Non-sliding horizon
Sliding horizon Planning built at t = 0 Planning built at t = P Planning built at t = 2P Planning built at t = 3P Planning built at t = H
0
P
2P
Revision
3P
4P =H
H+P
0
P
2P
3P
4P =H
H+P
New forecast
The GRAI method, Part 2: detailed modelling and methodological issues
50
«A» rule Vision of the future Non-sliding horizon
Sliding horizon
4P = H 3P = H-P 2P P 0 0
P
2P
3P
4P =H
H+P
0
P
2P
3P
The GRAI method, Part 2: detailed modelling and methodological issues
4P =H
H+P
51
«A» rule The control systems we take into account are based on forecasts (that is why the notion of horizon exists); The only way to permanently keep a minimal vision of the future is to have a sliding horizon; With a sliding horizon: Vision at decision period = H, Minimal vision = H - P; Note: when H = P, the notion of sliding horizon has no sense anymore and we consider then that the situation corresponds to a non-sliding horizon. The GRAI method, Part 2: detailed modelling and methodological issues
52
«B» rule The horizon must be longer than the time to achieve physical activities of production controlled by the decision centres of the level. L = delivery date Decision Decision The planning realized at L-d centre centre must take the result expected at L into account, then a T= Controlled vision in the future over d is Controlled T = necessary: L activity L-d activity ⇒H≥d Duration = d Note : This is true when there is a commitment about the delivery date (generally the case in production management). The GRAI method, Part 2: detailed modelling and methodological issues
53
« B » rule Definition of the relationships between physical activities of production and decision centres Knowledge coming up from production management
Master Production Schedule
Horizon ≥ D
Load planning
Horizon ≥ d
Production Purchasing
Manufacturing d
D
The GRAI method, Part 2: detailed modelling and methodological issues
54
« B » rule Precision The former general rule is true for the nominal running since the vision in the future is equal to H at the period of decision making. In the case where many adjustments are needed, the minimal vision in the future must be considered: H - P. Then, if the environment of the system is very disturbed, it is necessary that: H-P≥d The GRAI method, Part 2: detailed modelling and methodological issues
55
«C» rule
Ma S na tab ge ilit me y a nt nd lig htn e
ss ne ve
Stability: by implementing « enough » a plan before revising it. Management lightness: less effort with a batch-process of events.
ti ac Re
• Shorter the period is, more the system is reactive, • Longer the period is, more the system is stable and light to manage.
ss
The value of the period is linked to the frequency of fluctuations impacting the decision centre considered.
Value of the frequency
The GRAI method, Part 2: detailed modelling and methodological issues
56
«D» rule The ideal number of decisional levels is between three and five. Too much levels make the system too complex and is often the symptom of a synchronization issue between decision centres. Not enough levels does not allow a « progressive coordination » of the system. Empirical rule defined by experience. The GRAI method, Part 2: detailed modelling and methodological issues
57
«E» rule The horizon of a level must have a value equal or greater than the value of the period of the upper level. Sources of variations: Decision framework
Order
Decision centre
Follow -up
Orders and follow-up are continuously received, the decision framework is emitted only at each period of the upper level. Then, the environment of the decision centre is stable from this point of view during Pn+1 (period of the upper level). This stability is taken into account by having: Hn ≥ Pn+1 Empirical rule : Hn = 2.Pn+1
The GRAI method, Part 2: detailed modelling and methodological issues
58
«F» rule A decision framework must not jump a level. Configuration that does not match the rule:
• This configuration denies the raison d’être of the jumped decision centre and the interest for having a progressive co-ordination. • The decision centre receptor receives a frame not often enough and too global related to the detail it processes itself.
The GRAI method, Part 2: detailed modelling and methodological issues
59
«G» rule A decision centre must receive only one decision framework. • The objectives sent risk to be contradictory; • The decision space of the decision centre receptor corresponds to the intersection of decision spaces defined by each decision framework (decision variables and constraints). This space risks to be empty.
The GRAI method, Part 2: detailed modelling and methodological issues
60
« G » rule Decision framework
Configuration that does not match the rule:
Decision framework
Decision centre
Configuration theoretically possible (mainly if it exists a unique decision centre upstream the decision frameworks) but not practically recommended.
The GRAI method, Part 2: detailed modelling and methodological issues
61
«H» rule A decision framework inter-functions is possible only if the decision centre emitter belongs to a function whose the basic elements contain the basic elements of the function that the decision centre receptor belongs to. Reminder:
Activity control Production activity control: basic concepts Activity Activity management management Product Product management management PxT
PxRxT Resource Resource management management
Product Product
Activity Activity
Product Product
RxT
PxR P
Resource Resource R Date
«Activities management» ? «Planning»
Title
The GRAI method, Part 2: detailed modelling and methodological issues
26
62
« H » rule Example of forbidden configuration: To manage To plan products
The decision centre emitter does not master the notion of Resource: it cannot completely frame a decision centre belonging to the “Plan” function
The GRAI method, Part 2: detailed modelling and methodological issues
63
« H » rule
Possible configurations (for the three elementary control functions):
Theoretically possible configuration but forbidden by the G rule:
To manage products
To manage products
To plan
To manage resources
To plan
To manage resources
The GRAI method, Part 2: detailed modelling and methodological issues
64
«I» rule A function is an exclusive set of activities having a role participating to a common and identified finality. • Common understanding of the function and its finality, • Each decision centre of the function participates effectively to the finality of the function, • Consistent set of objectives deployed all along the hierarchy.
The GRAI method, Part 2: detailed modelling and methodological issues
65
«J» rule Each elementary control function must have a decision centre at each hierarchical level. Reminder: Activity control Production activity control: basic concepts Activity Activity management management Product Product management management PxT
PxRxT Resource Resource management management
Product Product
Activity Activity
Product Product
RxT
PxR P
Resource Resource R Date
Title
«Activities management» ? «Planning» 26
The «Activity management (planning) / Products management / Resources management» triplet cannot be dissociated ⇒ If one element of the triplet is present at one level, then the two other ones must be present as well.
The GRAI method, Part 2: detailed modelling and methodological issues
66
«K» rule The production management system must be informed enough about its environment (external information) and about the physical system (internal information). The system is: • open: information coming from its environment, • looped : information coming from the physical system. Knowledge about production management is necessary to know what information is required and where. The GRAI method, Part 2: detailed modelling and methodological issues
67
Conclusions about inconsistencies rules • Some rules are general and express control principles (GRAI model): they are applied to the grid (rules presented here); • Other more accurate rules come from production management principles and are applicable to the grid and to the nets (rules not presented here); • It is possible to enrich this set of rules (with rules dedicated to a class of systems, etc.). The GRAI method, Part 2: detailed modelling and methodological issues
68
Fifth
THE GRAI part METHODOLOGY
The GRAI method, Part 2: detailed modelling and methodological issues
69
The GRAI methodology: definition • The GRAI Methodology applies in the same general optics as the GRAI method (improvement of performances). • The GRAI Methodology is also based on a reference model, graphic languages and a structured approach. • The difference concerns: • the modelling domain, • the offer of specific approaches according to the aims of the study. The GRAI method, Part 2: detailed modelling and methodological issues
70
Modelling domain (by sub-systems) Entity/Relationship or class diagram
functions
entity
relation cardinality entity
DECISIONAL SYSTEM
H/P
INFORMATION SYSTEM PHYSICAL SYSTEM
trigger
support
activity
entity
GRAI Grid
support
result
control activity support
control activity
GRAI nets
support
Actigrams The GRAI method, Part 2: detailed modelling and methodological issues
71
Modelling domain: global representation Actigram control
FUNCTIONAL VIEW
activity
control
support
activity support
DECISIONAL SYSTEM INFORMATION SYSTEM control
extended activity resources
logical operator process
PHYSICAL SYSTEM
PROCESS VIEW
extended Activity
resources
Extended Actigram The GRAI method, Part 2: detailed modelling and methodological issues
72
The offer of specific approaches CHOICE AND IMPLEMENTATION OF IT/TECHNICAL/ORGANISATIONAL SOLUTIONS GIMSOFT Choice of IT solutions
PROGRAI
INDUSTRIAL SYSTEMS ENGINEERING
PERFORMANCE MANAGEMENT
GIMPLANT Implementation of solutions
Enterprise process improvement
ECOGRAI
INDUSTRIAL STRATEGY
Performance indicators
BENCHGRAI Benchmarking
GEM
Evolution management enterprise
GRAI Engineering
Product design control
GIM Audit GIM
Enterprise modelling
GRAI Model
Languages
GRAI Message Manufacturing Strategic Plan
GRAI Quality GRAI Knowledge
KNOWLEDGE MANAGEMENT
Generic approach
The GRAI method, Part 2: detailed modelling and methodological issues
73
Specific approaches and life cycle Evolution control oriented
SIS oriented
Definition of objectives and strategic planning
GRAI KNOWLEDGE
GEM
GRAI QUALITY
GRAI MESSAGE
GIM PLANT
Implementation of technical solutions ECO GRAI
Exploitation/follow up of implemented system
BENCHGRAI
GIM SOFT
Choice/development of technical solutions
GRAI ENGINEERING
Target system design
GIM
Diagnostic
GIM AUDIT
Modelling of the existing
The GRAI method, Part 2: detailed modelling and methodological issues
74
Thank you for your attention
The GRAI method, Part 2: detailed modelling and methodological issues
75