A project plan is a tool that enables the project management to allocate resources and attention appropriate to the status of the project. When I mention the “status of the project” I mean specifically whether the project is late or early, whether the project is running at high efficiency (earned value) or low efficiency, and so on, in accordance with the latest update of the plan. The project plan also enables the project management to hone in to the specific tasks (i.e. those on the project critical path, or those that are running at low efficiency) that the program reveals require attention.
It goes therefore without say that if the duration of any or many an activity is wrong, the usefulness of the plan is seriously, if not fatally, impaired – the criticality of any activity becomes questionable with the loss of integrity of the critical path, the efficiency measure becomes questionable [even in relationship – unless the error in duration is a constant factor], and the earned value relationships become unworkable (because the relative work or time content of each activity is not in reality even relationally as reflected .in the baseline plan).
To arrive at an estimate of durations one or all of the methods described below are generally applied.
1. Work content estimating (Low-level [level 4+])
The estimate of duration of any activity on a low level detail plan is generally accompanied as an integral part of the measure by an analysis of the work content of the activity (on a level four or greater level of detail schedule).
At this level of detail (levl4+) most multi-action activities are broken down into unique actions (modularised) to allow the estimator to arrive at his base activity work content estimate.
An example would be a “weld section a to section b” activity – this would be broken down into possibly the following actions:
- collect section a;
- collect section b;
- set-up section a and b;
- weld prep section a and b;
- set up weld gear;
- pre-heat weld section;
- post weld heat treat;
- Register for NDE
Each of these activities would have a man-hour and duration applied. The duration would be based upon the estimator or the work supervisor experience. Because the activities are broken down to a base level of experience, the durations of the actions are known with a great deal of certainty and
2. Use of deterministic data
Part 1 – “Norms” (Level 4+)
In many organisations there is available in-house a collection of data relating to standard durations for low-level activity operations.
In several instances professional estimators have access to industry standard norms (performed on the North Sea oil rig/platform projects)
To arrive at the work content and duration for a planned activity the planned activity is broken down into actions and the norms are applied to those actions.
The level of breakdown or modularisation of the activity would be dependant upon the level of detail of the “norms” that the estimator was applying.
The total activity durations that are established by applying the modularisation and application of norms are then modified by socio-economic and physical parameters peculiar to the project and specific task (e.g. a percentage increase for lower skill levels, a % increase for field work as opposed to site work, a % adjustment for environmental conditions, a % adjustment for limited access, whether it is winter or summer, whether it is the rainy season or the dry season, etc. etc.) and sociological impacts (holiday season, religious festival season, elections, strike season, etc). At the end of the day, however, the durations estimated can be qualified [because the norms are deterministic (i.e. have been arrived at by physical measure and/or actual data collection and average) and any difference can be related to the quantification and application of adjustment factors.
The “factorisation” of norms-based estimating requires a high level of skill and experience, and should not be attempted by novice estimators or planning engineers.
This level of work content based duration estimation is only available to a project team once the detail of the work to be done has been settled [generally some time after completion of detail design] and is therefore of application only to the Contractor (rarely) or the subcontractor or sub-subcontractor on a macro (US$ 150 million+) project
Part 2 “Norms” High level – Level 3 to level 1
On an industry, corporate and personal experience level there are a large number of cost and time related relationships that are clearly apparent at a higher level of planning.
These norms are distinguished from “estimating norms” by level of activity [estimating norms being low level, other norms being higher level]. For want of a better term I call them “Cost Engineering Norms” where they relate to relative values/resources, and “Planning Norms” where they relate to durations – fairly often the two are closely interrelated.
For example, “Engineering and Project management” would be 10% of the total cost of a certain type of Greenfield project; “Design and Drawing” would be a 12 month overall duration for a specific type of project. These norms are the basis for “sanity test” of the cumulation of level 4+ activity resources and durations.
3. Use of Stochastic [sampling and probability distribution] methodologies
Part 1 – Triangular analysis using Monte Carlo sampling
Estimating is not an exact science. The word “estimate” in fact means to approximate and is almost the opposite of “exact”.
As noted in the introduction above, however, a large degree of accuracy in duration and resource levels is required if the appropriate use is to be made of formal planning methodologies and algorithms.
A method of approaching this level of accuracy involves the objective evaluation of the data prepared by the planner/planning team by the appropriate members of the project execution team.
This is a modification (I believe a practical and effective modification) of PERT application.
For each activity duration, the appropriate PM team member (contract engineer, construction engineer, construction manager, foreman (if available), section leader, etc.) will stipulate, within the context of the “planned” duration [and this assumes that the most likely duration is the planned duration – if not, then the planned duration should be re-examined], the “best” and “worst” durations and the probabilities of either of these occurring (best % + most likely % + worst % = 100).
Once this has been carried out (generally at a comparatively high level of activities – say not more than 250 activities at a time), aMonte Carlosimulation of the durations and the impact of the change probabilities is run. This produces a probability distribution curve that can be related to the achievement of any particular project milestone or to the overall project completion date (the same principles can of course be equally applied to resourcing).
As a sanity test the peak of the distribution curve should be (however, does not have to be) approximately the same as the “planned” completion date.
The probability of achievement of any particular completion by any specific time can then be read off the curve (time on the horizontal axis and probability being plotted on the vertical axis).
Cumulative probability is then purely a cumulation exercise.
The exposure of the project to time risk can then be easily demonstrated and can be directly addressed by asking the PM team member what information/commitment they require to bring “best” and “worst” closer to “most likely”.
The exercise should be carried out at key intervals in the project execution phase and the probability distributions of each successive distribution, if plotted on a trend line, should show progressively lower levels of risk in completion date forecast.