| |
Project scheduling: out of stall
The project activity plan may contain 2000 - 6000 entries for seawater desalination
project of 100,000 - 300,000 m3/day. Maintenance of such big plans requires
2 - 4 managers and the annual budget of US$ 80,000 - 200,000.
The only tools in current use for project scheduling and tracking are
Microsoft Project Office and its functional twin -
PSnext.
Both are built around the Gantt chart and have a wealth of amazing features.
(PERT diagram offered together with the Gantt chart is not applicable
to the construction projects like the desalination ones and so is not
considered here.)
The author’s survey of projects already done and under construction
identifies the following Gantt-chart-related problems.
-
Interdependencies between activities and milestones are not always
shown - logic of activities is broken (in 40 - 70% activities).
-
Heavy-weight activities like ‘Update of...’ and ‘Correction
of...’ become normal practice in the environment with broken logic.
-
Resources are not detailed (in up to 80% activities).
-
Decisions to ‘outsource’ part of the time-consuming
work (mostly drafting, up to 40%) due to resources scarcity can not
be deduced from the Gantt chart analysis.
-
Even for typical activities (P&ID drafting, equipment lists and
instrument lists preparation) durations instead of work loads are used
(100% activities).
-
Activities lasting less than 1 week are not filtered out (25 - 50%),
the project schedule is turned into the ‘todo’ list.
-
Building the Gantt chart even for grass-root project takes many hours
- roughly 1/3 hour per activity.
-
As Gantt charts are mainly duration-oriented, updates are time-and labor- intensive.
-
Updates are always late especially those reflecting the project scope
change and new requirements by client (from 1 to 4 weeks).
-
There is no possibility to implement “what if” scenarios
when, for example, testing and/or inspections are poorly done.
-
Project data summary and scheduling optimization (critical path analysis,
resources leveling and allocation, slack time minimization, probability
analysis, prediction correction, etc,…) are not performed routinely
for example, on a weekly basis).
-
Scheduling risk analysis & assessment of non-typical solutions
are not implemented.
-
Assignees are not involved in the progress reporting on the daily
basis.
-
Managers are not qualified enough to make sound predictions of the
project would-be status and what specifically should be done to improve
the situation.
These drawbacks repercussion on the project management is so heavy (cost
overruns, schedule slippage) that one may doubt whether such high-quality
software like Microsoft Project Office or PSnext is applicable at all to
mega-projects with more than 200 - 500 activities.
This situation is further aggravated by the fact that even in the largest
companies years of experience in the scheduling techniques are needed (mainly,
to collect historical data).
There exists a distinct contrast between planners and doers. During planning
phase the latter always add some “fat” to their biased estimates
of the work load. (By the author survey the range of estimates given by
managers and engineers may differ by a factor of 4!) Once the schedule is
constructed the planners become observers and rely on the doers to accomplish
the job within time limitations. Management concentrates its effort on convincing
(!) the doers that they have obligation to the successful completion of
the plan. In other words, difficult chore of the plan ongoing update often
has no addressee.
Sometimes the clients require that formal project plans and schedules be
provided thirty to ninety days after contract award. (For comparison, preparation
of the schedule for the high pressure pump of 2000 m3/h takes approximately
30 days) As a rule this activity is supposed to be done in parallel with
the P&ID drafting and the equipment lists preparation – the main
source of the project scheduling information. This latter work package requires
from 3 to 6 months. The time mismatch is obvious.
Adopted scheduling techniques are mostly focused on what to be done irregardless
of nature of work. The example below explains this point in more detail.
In a large desalination project the novel design of the flap valve for dual
media filter was suggested. In making decision to go with this design, the
project manager overlooked 2 important points.
-
The company engineers were not unanimous about the valve workability.
-
The order was awarded to the company having no experience in design,
manufacturing and servicing of such kind of valves.
Worst-case scenario surfaced - the plant construction was paralyzed for
nearly a month before commissioning: defective valves were re-designed and
repaired. Total cost overruns (indirect, material and labor, lost opportunity)
were about US$ 4.5 million.
Towards scheduling automation
The Piman project manager (PPM) has been developed with the following tasks in mind.
- Project logic planning to meet the project delivery scope,
- Generic resource planning to meet the project time frame,
- Cash inflow auto-generation,
- Project progress tracking.
In the project planning PPM is 100 times faster(!) than any of the above-mentioned
programs. Another big step forward is generic resource optimization - starting
point for any project, that is never done in practice. PPM uses activity-type
network depicting the activities as arrows and events as nodes. This network
is referred to as Arrow Diagram Method (ADM) or Critical Path Method (CPM).
Activity is considered an atomic and indivisible piece of work that may
be done in isolation from the other activities.
PPM has the following specific features.
1. Graph math
Network diagram is mathematically described by a unidirectional graph theory offering
rich set of analysis techniques.
2. Activity interdependencies validation
PPM validates that the network diagram have always a single start and single
end points and neither activities nor events are dangling. This limitation
prohibits merging of the project with R&D projects started earlier. Such
merger has more than one starting point.
3. Subgraph structures
An activity may be a simple task or a multi-task containing another network
diagram - structured data - or an ordered list of unstructured activities.
This allows easy integration of the subcontractor schedules and “what
if” scenarios.
4. Customized activities keyed into P&ID
PPM takes advantage of the fact that most project activities are rooted into
or may be traced to the P&ID data. Any activity content description
is represented by two parts - description of deliveries and their scope.
For example, “assembly drawings of pretreatment area” may be
split into “assembly drawings” and the project sub-area called
“pretreatment area”.
PPM offers a static library of standard deliveries (PINs, orders, P&ID
items and approvals, interlocks and control loops, auditing results, engineering
reports and certified drawings) valid for any project and a dynamic library
of the scope entities like the project area, equipment class, supply packages,
P&ID diagram and others. This library is repeatedly auto-generated from
the project database for the schedule update purposes. Any activity may
be easily compiled from these two libraries.
5. Outstanding issue tracking context
Analysis of the hard-to-customize activities shows that most of them are
"light" or "todo" activities requiring less then 50
hours of work (one week). With Piman this body of activities is moved under
the roof of the outstanding issues tracking manager. To implement this the
standard activity has been turned into an entry to the database of
outstanding issues.
If some issue is not resolved the activity approval is blocked. This approach
drastically reduces (by 20 - 40%) the total number of activities engaged
in the project schedule.
6. Predictor - calculator method
Activity weight (represented as duration or workload) is proportional to
the activity scope, the activity type (assembly drawings in the example
above) defining the proportionality law. Here the activity type is predictor,
the activity scope - calculator. The predicting method depends upon the
project phase. For example, if the "Instrument index" activity
is done, the workload of activity called "IO list" is a product
of the number of instrument index list entries and the entry weight of 0.1
hour. At an early phase of the project engineering when the instrument index
list is not ready another correlation based on the project size is used.
7. Schedule template library
PPM auto-generates the network diagram from typical templates like "Turn-key
project" template or "Docs & order project" one. These
templates may be modified - unnecessary work packages can be easily removed
from the project.
Templates automatically incorporate optimal path patterns –
activity sequences reflecting best practices adopted in the project execution.
For example, “equipment layering” (EL) pattern defines the order
of the equipment sizing – heavy equipment with the longest delivery
times should be sized first.
So instead of “go-with-stream” sizing of all items on the P&ID
diagram and moving to the next one, only heavy equipment shall be sized
on all P&ID diagrams. Then less heavy “layer” shall be sized
and so on.
The following order of layers may be recommended.
- large pump sets of 500 kW and above,
- vessels and tanks,
- valves of ANSI class #300 and #600,
- filters and membrane arrays,
- all piping and balance of valves,
- electrical equipment – variable speed drives, MCC panels and
switchgears,
- auxiliary turnkey packages,
- instruments,
- all other equipment
EL pattern is also applied to the piping procurement: metal piping and fittings
shall be ordered first, and then the FRP piping and fittings followed by
the plastic ones.
8. Delivery time predictor
It is often said that the desalination project is all about delivery times
(DT).
They may be categorized on DT for the final product and DT for the engineering
documentation pack. The latter usually includes the general assembly drawings
(GA), cross section ones (CS), loads and forces distribution, the manufacturing
plan, the inspections and tests plan and others. The documentation pack
DT is a starting point for any detailed mechanical design while the product
DT is used for the plant assembly planning.
PPM calculates expected delivery times for typical equipment classes based
on the equipment type, size, and the design and construction attributes.
9. Generic resource analysis
The generic resource information generated by PPM is sufficient to perform the
following tasks.
-
to establish the optimal ratio between different resource categories
like process, electrical, instrumentation and control, mechanical and
others;
-
to define the resource quantity necessary to meet the project deadline;
-
to define the best strategy in case of insufficient resources - to
hire or to outsource;
- to define the resource availability for other projects;
- to predict the resource costs against time distribution;
- to define the company capacity usage in the long-term activity planning;
The sample chart to the right is auto-generated by Piman and shows the effect
of optimal resource availability on the project time frame. Any project
manager wishes he or she had this chart.
The test runs of PPM immediately produced results which are in excellent
agreement with current management practice.
- Desalination project scheduling is mostly defined (75%) by the equipment
delivery times.
- "From-scratch" turn-key projects cannot be completed in
less than 2.7 years, and only "copy" projects hit the limit
of 2 years.
- The resource average usage in a project is inherently low, and the
work outsourcing should be a common practice. The electrical, instrumentation
and control, mechanical and civil engineering are best candidates for
outsourcing.
- The heaviest part of the project is mechanical design which may be
executed by any engineering company even having no previous experience
in the water desalination.
10. Resource auto-naming
To replace generic resource with named one in the programs like MS project office requires at least
4 steps and a lot of brainwork.
Such a chore for mid-size desalination project with more than 500 resource-linked activities
may take 2-4 weeks. To the best of my knowledge this replacement is never
fully done in practice. Without named resources the project progress forecasting
- the basis for the project success - is not possible.
To automate naming resources, Piman tracks the user activities across the projects
and auto-creates the User Resume. The rest is straightforward.
[to be continued soon]
| |