Three standard operators for contribution to a recursive planning process
Pr Jean-Claude Bocquet, Ph.D., Industrial Engineering Laboratory, Ecole Centrale Paris, France Franck Marle, Industrial Engineering Laboratory, Ecole Centrale Paris, France
Historically, project organization has been done, first with the parameter time (Gantt charts, program evaluation and review technique [PERT], schedules), then with the parameter hierarchy (work breakdown structure [WBS]). But many other aspects of project management are not present in these tools. This could explain a statistic extracted from a survey done by the Standish Group International on several hundreds of information technology (IT) projects: more than 40 percent of projects do not compare real data to planned data, and 43 percent do not update their plans and schedules. The fact is that people feel that a WBS or a Gantt chart are not enough to completely describe the project and its complexity.
Concepts of project, deliverable, activity, and objective are confused. Some people can speak from ones by thinking to others, and others can put on the same diagram several of these objects. For example, the terms target, purpose, goal, aim, and objective may be used at different times or in different situations. Another example is a small project of reservation of a plane ticket. The objective is to have the ticket plane booked. The deliverable is the ticket plane. The activity is to book the ticket plane. It is easy to have confusion.
For resources, distinction is not well done between responsibility and execution. Skill requirement is a problem, because it is already difficult to know which skills are needed. The kind of competence required for an executing person or for a responsible person are not the same, in terms of nature and value.
Then, the tools today tend to optimize a global situation, which requires having the complete situation described. For example, resource assignment optimization requires having the complete list of activities. And this is not easy or even pertinent to get at the beginning of the project.
Research problematic is then to:
• Make an effort to structure and define the 7 objects that are part of a project or are around a project, like deliverables, objectives, activities, or actors. The purpose is to repeat the structuring effort done for the activities to the other objects.
• Be able to describe the complete project at each time. This will be recursively done by using 3 standard operators. They will allow going to the detail level which is required at each time. This will be done as soon as the project has officially begun, with the manipulation of the previously introduced objects with the operators of decomposition, assignment, and advancement.
• Get at each detail level at each time and for each kind of object, more information than the only hierarchical and sequential links. The whole interactions of an object with its direct environment overwhelmed the hierarchy (WBS) and sequentiality (Gantt, PERT). A kind of magnifying glass will be used to make visible this additional information, which is not visible with traditional tools. The complexity of a project makes impossible to visualize simultaneously the complete project (global vision) and the whole interactions in the project.
This article introduces details about the first two points, and some notions about the third. Research is applied by a French automotive fabricant, PSA Peugeot-Citroën. Application concerns the whole three research topics, and consists of the development of software that helps to project elaboration by:
• Additional structuring which distinguishes the objects
• Operators which help the three fundamental project elaboration actions
• Making available a visual representation of all the information linked with an object.
The 7 Objects
A list of 7 elements that constitute a project or interact with a project has been done:
• Project: the basic element, the project itself. The Project Management Institute (PMI®) definition is “temporary endeavor undertaken to create a unique product/service.” The project itself is unique by the uniqueness of each situation, described by the resources, the environment, and the method. So, even if the objectives are the same, the project will be different.
Exhibit 1. Hierarchies for Several Natures of Objects, and Links between These Hierarchies
• Deliverable: measurable and tangible result or item that must be produced to complete project or part of the project. It can also be defined as a “concrete representation of an intermediary or final result of the project.” The important notion here is concrete, tangible. A deliverable is something material.
• Objective: “vision of a future situation to reach, or a future desired situation.” An objective is something to reach, but is not material.
• Activity: “succession of elementary actions realized by resources, consuming time and money, and producing some results contributing to some objectives.” Deliverables are realized by activities.
• Decision: choice made between some identified potential solutions. Solutions can have been analyzed, estimated, quantified, and validated, or not.
• Actor: human resource or human structure (like a company, a subcontractor) that can realize activities and make decisions. Actor may be assigned to activities (executing resources), or to project (project manager), or may have an influence or may be influenced by the project results (stakeholders, shareholders, management, customers).
• Process activity: restricted to service projects, it describes the impact of the project on the process. The project modifies the process of the company, so the new process is defined by transformation of some of its activities. This object can be or not in the scope of the project: for a project of transformation of an internal process, it is in the scope. For a product development project, it is not.
These objects are not clearly defined today, especially in planning tools like WBS or Gantt charts. The proposal is that hierarchies can be built with only one nature of object, and links between hierarchies are to be created when necessary (see Exhibit 1). For example, WBS is theoretically the deliverables hierarchy; there can be an objective hierarchy and an activity hierarchy. As an activity contributes to a deliverable, this creates a link between the both hierarchies. The same thing works between deliverable and objective.
The list has been built with the following question: what are the existing interactions in a project, and between what objects do they exist? This allowed building simultaneously the list of objects and a list of types of interactions. The list of interactions is not in the scope of this paper; it will just be introduced in the decomposition part.
How to be sure this is the right list of objects? One can never be sure, but two things are sure:
• All the listed elements have a reason to be there, so the 7 elements are justified, because at least a real-life example was found for each.
• If there is another object that had been forgotten, it could be added to the list without any problem. Further research will enlighten the risk case: can an event be an object?
This part introduced 7 types of objects that describe the project and its interactions inside and outside. Next part presents how 3 operators can manipulate these objects, in order to elaborate and make progress the project. The operators are standard, which means they can be applied at each level of the project and for each object. The first one is decomposition, which builds hierarchies. The second one is assignment, which makes concretely feasible immaterial objects like deliverables or activities. The third one is advancement, which tracks the situation of the object on its way to the final state.
The 3 Standard Operators
Each operator is a decision process. It is a choice between several alternatives, in which can be the alternative of doing nothing. The list of 3 operators described previously has been defined with a 4-step decision process:
• Generation: the actions required to create new or find existing potential solutions.
• Evaluation: the actions required to select and sort the potential solutions, by measuring some choice criteria. The decision is taken at the end of this step.
Exhibit 2. Global Scheme of a Decomposition
• Application: the actions required to make the decision concrete. The operator is executed. Objects are manipulated and transformed.
• Storage: the actions required to keep useful information about the past, to be able to reuse it next time.
Decomposition is a decision that cannot be automatically performed. It is the operation of expressing an object into several smaller ones. It is a complex decision that requires human thinking and feeling.
The nature of the object may be the same (a deliverable into smaller deliverables), or not (a deliverable into activities, a project into deliverables). It gives two natures of arborescence: the hierarchical one, with one nature of object, and the contributing one, with two natures and only one level, as it was introduced in the objects part (see Exhibit 1).
Potential solutions generation is based on historic and creativity: it needs historical information storage and retrieval tools, and creativity or systematic analysis methods.
A particular method can be used to generate a decomposition, and this is a research result: a decomposition is always a linear combination of mono-criteria decompositions, the criteria corresponding with the existing interactions between objects.
This list of interactions is just introduced here:
• Hierarchical (father/son, like in WBS)
• Contribution to (a deliverable contributes to an objective)
• Functional (similarities between two objects, that allow to work together or get some best practice)
• Sequential (like in sequence diagrams or Gantt charts)
• Influence (stakeholders that can have an influence, like management or sponsor)
• Resource (similarities between assigned resources or required competence)
• Exchange (exchange of data without link of contribution or influence).
This axiom will be developed in a further research.
Evaluation of the potential solution is helped by checklists, with 7 categories of questions. For example, category 1 is about completeness of the solution. Questions are: “If the objects sons are realized and their objectives reached, is the object father realized and its objectives reached?” “Are there some things in the objects sons that contribute to nothing in the object father?” “If some things are missing, is this possible to express them into a or several new object(s) son(s)?” “Do the objects sons have delimited frontiers?” “Is there some recovering between the objects sons, and if yes, are they volunteer?”
Exhibit 3. The Two Resources Spaces
Application of the decision consists of two points: communication and control.
The whole persons and structures that are concerned by the decomposition decision must be informed: the “grandfather” (the father of the object father), the “uncles” (the brothers of the object father), all the objects linked with at least one of the objects sons.
Decomposition has the advantage of creating smaller, more manageable objects. But it requires a coordination effort and an aggregation that must be taken into account. The feedback loop must be created as soon as the sons are created, and regularly followed.
Storage is about useful information, information about alternatives (why were they not chosen?) and about results (did the decomposition work? shall the solution be reused?).
Exhibit 4. The Global Scheme of an Assignment Decision
Assignment is a decision. It is the operation of putting material resources, like persons, tools, equipment, or machines, to immaterial objects, like a deliverable or an activity. An activity for example is nothing more than a concept. It needs human or mechanical energy to be concretely executed.
Assignment may be done with several types of resources. Scope is limited to human resources. There are only two categories of actors, related to two types of assignment: responsibility and execution. A person who is assigned to an object has to do something, or to make something done by someone else.
• The responsibility assignment is for project, deliverables, and objectives: a project manager is responsible for the achievement of the project.
• The execution assignment is for activities and decisions. People execute some elementary actions in a certain sequence, which allows executing the complete activity.
The last two types of objects are not relevant with an assignment.
The process will be the same for the two natures of assignment, but there is a difference in the expression of the competence requirements. Indeed, if a person is responsible for something, she needs enough competence in order to make the work done. But it does not mean that she is able to do the work herself. She has enough technical background, with some management capacities. A person who has to execute an activity will have competence required to do the work. But it may be someone without any skill in management.
Resources may be separated into two categories (see Exhibit 3):
• The material resources: they are something material, tangible, like persons, machines, equipments, or tools. Material resources can execute activities, and in the case of human resources, make decisions.
• The immaterial resources: they are required, but not tangible, like competence or technologies. They indirectly contribute to actions and decisions. They are made too of knowledge, experience, personality, and other human characteristics.
Using a material resource consumes it, in terms of time, money, or capacity. Using an immaterial resource does not consume it; it can even increase it (knowledge, skill).
There is a correspondence between these two resource spaces. Competence corresponds to persons, and technologies to machines. A person can be described by her material characteristics (cost, size, age) and by her immaterial characteristics, which are in the immaterial resources space.
The assignment process itself is simple: it is required to travel through the immaterial space until the material space. In order to find an adequate material resource, it is necessary to first define the required immaterial resources. They will be used as criteria for the choice of the assigned resource.
Assignment is an iterative process, which needs several in and out travels between the three stakeholders: the person who wants to assign, the person who is susceptible to be assigned, and its hierarchical manager.
It begins with the identification of the required skills and competence. Then, the persons who have these characteristics are selected. As several persons are selected by this elimination process, it needs a sort, to be able to choose.
The criteria have to be identified, and measured if possible. A first selection can be partially done with the available data. A second one has to be done after the completion or checking of some data, like availability, interest and some human qualities, characteristics. The interest of the person for the potential assignment has to be identified, and is as much as important as the technical and managerial competence.
The decision is taken with this additional information, and with other criteria that will never be modelled, because they are unconscious. That is out the scope.
The decision needs to be applied, by updating the information systems and by informing all the stakeholders of this decision, not the whole project. Every person concerned by the assignment needs to be informed.
Then, useful information may be kept for future reuse in similar cases. Information in assignment is more useful about immaterial resources than material resources, because the material resources change. They arrive and go away, so are less stable. The choice itself is sometimes less important that the justification of the choice: what were the criteria, how reliable were the data, what made the difference for final decision, …
Exhibit 5. The 3-Step Advancement Model, Introducing the Preparation Phase
It is characterized by an advancement state, which can be placed on an oriented scale made of several intermediary states:
• Estimated: the object is created. Some data and information are not available, and do not allow to take decisions, like when, where, how much, who, and so on
• Planned: decisions have been taken. The future of the object is planned, it needs to be realized. Some resources need to be booked, and some contracts/procurement need to be made. There is already some irreversibility, because some time and some money have been consumed.
• Under development: execution is ongoing. There is more time and more money consumed. The work is in progress, and new decisions may have consequences on work already done.
• Done: the work is done and validated by the appropriate person(s).
The decision is to put the advancement state from one to the following. Scope is about the date of the decision. Indeed, a too early decision involves the risk not to have the right and reliable information. Decision may be bad, and may involve mistakes, rework, and penalties. On the other side, a too late decision involves the risk of the activity itself to be late.
The risks are not symmetrical: there is the risk of a bad decision (too early decision) and the risk to be late (too late decision).
A solution to fight these risks is to put on the visible side the preparation phase, between the planning decision and the beginning of the execution (see Exhibit 5). Preparation is the whole actions required to book resources and to manage contracts, procurement and the whole logistics. It prepares the action, the battle. The generals plan the battle, and the soldiers fight, but between there are a lot of things to do. If these are forgotten or not well cost and time managed, battle can be lost.
A research proposal is to determine an estimation formula of an optimal date of decision for advancement state change. Optimal is a very bad word, which does not mean what it really is: an estimation with incertitude in the calculation and in the interpretation.
As the responsible person knows the end date of the activity (for example), knows its duration, and estimates the preparation phase duration, she can have an estimation of the decision date not to go over:
Dd = End date – Execution duration – Preparation duration
This must take into account all the incertitude about durations and dates.
Detailed formulas are available today, but not developed in this paper.
This allows calculating an estimated late, or an estimated probability of the activity to be late. As probability and impact (the number of days late) are known, it is equivalent to a risk management process, with the notion of gravity. It is a risk of an activity to be late, due to a decision taken too late.
Last thing is about simultaneous independent management of the activities advancement state. Classical tools make the whole activities go from estimated to planned state in one shot. The rolling wave concept allows going from estimated to planned state with one fixed duration. The research proposal is to manage independently each activity, in order to adjust this duration to the individual case. The estimation of decision date described above allows adjusting the length of the wave for each activity.
How Can These 3 Operators Build a Project?
The operators build the project hierarchies, make the link with concrete resources, and make the project go through different states until the final one. It is enough to basically describe how a project lives and is described. Of course today more information are needed, but the heart is about these three phenomena.
Concepts are recursive, which means that they can be applied at each level of the project, in each hierarchy at each level of the hierarchy. The difference between a car development project and a screwdriver development project is the magnitude, the context, and maybe the pressure, but they are both projects. The car and the screwdriver are both deliverables. The concepts have different values, but are of the same nature. So, some operators can be applied independently of the magnitude, of the detail level of the object. Secondly, as objectives and deliverables need to be decomposed, why do not use the same operator? As activity and deliverable need resources to be executed, why do not use the same operator?
Industrial Application in a French Car Manufacturer: PSA Peugeot-Citroën
PSA is one of the two top car manufacturers in France. As every international company, it has numerous internal projects, which tend to transform internal processes and organizations. As PSA counted roundly 250 projects simultaneously, with different advancements and different management methods, it was decided to put some visibility.
Objectives are to get more visibility about what the projects transform and for whom, in order to be able to manage more globally and more consistently the whole 250 projects.
The research proposal is to build a tool in which three parts are identified:
• The central part, which is the object.
• The intermediary part, which describes 7 internal characteristics of the object. This point is developed, but not in the scope of the paper.
• The external part, which describes the 7 external interactions already introduced, but not developed in this paper.
As this tool is standard for each object, it allows to:
• Clarify what is what and avoid confusion between object natures.
• Clarify what is the situation of the object at each time, with its characteristics, in which assigned resources and advancement state are.
• Clarify with what and for what reason the object interacts, in which decomposed object sons are.
This is the magnifying glass briefly described in the introduction.
A software prototype is under development. The concepts are validated by industrial use. The next step is to change in depth some practices to get better performance and more standardization in the project management methods.
Conclusions and Perspectives
Some concepts have been introduced in order to clarify the definition of the objects and their interactions in a project, and outside a project. As objects are clear, it is possible to properly manipulate them, with only 3 decision operators, which allow building object hierarchies, with resources assignment and advancement state. As concepts are recursive, equivalent at each detail level and for each object nature, it is possible to use the same magnifying glass to visualize more information than the only hierarchical and sequential interactions. These three parts, objects, operators, and magnifying glass, are the key points of the industrial validation that are ongoing for ed 2001 to October 2002. The complexity of management of 250 projects simultaneously is reduced by the use of the magnifying glass, which shows more information, especially on the operators applied to well-defined objects. Estimation of the impact has already been done, but further research will be on quantification on the return on investment on such a research.
Earl, C., J. Johnson, and C. Eckert. (2001). Complexity in planning design processes. 13th International Conference on Engineering Design in Glasgow, Scotland.
Kloppenborg, T., and W. Opfer. (2000). Forty years of PM research: trends, interpretations and predictions. In D. P. Slevin, D. I. Cleland, and J. K. Pinto (Eds.), Proceedings of PMI Research Conference 2000: PM Research at the turn of the millenium (pp. 41–60). Newtown Square, PA: Project Management Institute.
Laurikkala, H. (2001). Reducing complexity of modelling in large delivery projects. 13th International Conference on Engineering Design in Glasgow, Scotland.
Le Moigne, Jean-Louis. (1990). la modélisation des systèmes complexes. Bordas.
Leroy, D. (2000). Project management practices in french organizations: A state of the art. In D. P. Slevin, D. I. Cleland, and J. K. Pinto (Eds.), Proceedings of PMI Research Conference 2000: PM Research at the turn of the millenium (pp. 339–346). Newtown Square, PA: Project Management Institute.
Morris, P. (2000). Researching the unanswered questions of project management. In D. P. Slevin, D. I. Cleland, and J. K. Pinto (Eds.), Proceedings of PMI Research Conference 2000: PM Research at the Turn of the Millenium (pp. 87–102). Newtown Square, PA: Project Management Institute.
Prigogine, Illya. (1996). De l’être au devenir.
Project Management Institute. (2000). A Guide to the Project Management Body of Knowledge (PMBOK® Guide) – 2000 Edition. Newtown Square, PA: Project Management Institute.
Simon, Herbert. (1991). Sciences des systèmes, sciences de l'artificiel. Dunod.