Transforming Construction?
6.0 Conventional Construction Management
In the construction phase, the absence of data problem arises
mainly in regard to project management or project control
activities: planning and scheduling and cost control and related
functions. The main problem here is with poor forecasting, giving
rise to unpredictable cost and schedule outcomes. Project forecasts
fail for two broad reasons: because the wrong targets are set at the
outset; and because progress is assessed incorrectly during the
course of the work. The way in which these and related problems
arise can be illustrated by focusing on their effects on project
planning.
6.0.1 Problems with Planning.
The conventional planning function involves the creation of
activity based project models, generally in the form of CPM networks
and Gantt charts. These models have two main purposes: initially to
demonstrate that the project at hand is physically / logically
achievable; and subsequently, during the course of the work, to
provide short-term guidance to action, in the form of look-ahead
schedules and such like. But activity models are flawed in both of
these areas of use, as the following will make clear.
6.0.2 Subjectivity
Activity planning models are highly subjective creations. On a
given project the particular activities chosen, the way in which
these activities are defined or specified and the logic links used
to connect them are all determined by the personal experience and
intuition of the individual planner or project manager. There is
nothing standardised or systematic about the activities. In a sense,
an activity means what the planner wants it to mean. This has three
highly undesirable consequences:
First, the sense or scope of a given activity tends to be
cryptic, difficult for other people to comprehend and therefore
difficult to challenge or verify independently. This undermines the
model's value as a test of the viability of the project. It also has
adverse consequences for the model's subsequent use as a management
control tool.
Second, the meaning of the activity is not determined and fixed.
It can be changed by the planner through the duration of the
project. This means that analyses and supposedly like for like
comparisons between different points in time on the project can be
misleading.
Third, activities tend to very project specific, reflecting the
individual planner's personal response to the project at hand. This
means that there is little scope for cross project comparison or
analysis which, again, makes it very difficult for people other than
the planner and his project manager to challenge the plan.
Also as a result of the unsystematic way in which they are used
to represent the scope of work of the project, it can often be
difficult to carry out progress assessments on a consistent, like
for like basis using activity models. There is often a tendency
towards optimism bias and other assessment errors which tend to
obscure the true situation. This makes performance trends difficult
to spot and effective responses hard to devise.
6.0.3 Planning versus Forecasting
It's important to distinguish between plans and forecasts. Once a
given project gets under way, the main requirement of the plan is to
provide guidance to action: rolling wave look-ahead programmes,
short term activity schedules and suchlike. To do this effectively,
the plan must reflect accurately the current, real world situation.
It must therefore be revised more or less continuously to embody the
effects of new information and changing circumstances. The only way
this can be done is by re-defining, or re-specifying the activities
that make up the model, or by re-arranging the logic links between
activities.
Once this starts to happen, like-for-like comparison of the model
at different points in time becomes impossible, and the ability to
generate consistent long term forecasts is lost. For this reason in
particular, planning models make inherently poor forecasting tools.
The planning model cannot, logically, be both a short term guide to
action and a reliable tool for consistent longer term forecasting.
6.0.4 The Top Down Problem
Conventional project management systems are essentially top down
in their structure and operation. It is extremely difficult to
integrate them closely with the detail of operations at the
production level of the project. This places an great deal of
reliance on the front line supervisor to act as the interface, the
interpreter, between the planning system and operations on the
ground. For the most part individual supervisors can be relied upon
reasonably well to understand the plan and to translate its contents
into detailed task schedules.
However, when assessing and reporting progress achieved, even the
most competent and most experienced supervisors are sometimes
betrayed by what academics describe as
Optimism Bias. This has been
defined as: "a cognitive predisposition found with most people to
judge future events in a more positive light than is warranted by
actual experience." In the present context it refers to the
intuitive reluctance of people to convey bad news to their
superiors; instead they delay acknowledging poor progress, often
until it's too late to correct. And this happens all the way along
the reporting line, from site level right up to the Main Board.
This reporting problem applies to all project management
disciplines and is evident at all management levels in project
organisations. Everyone involved interprets the information he's
given subjectively - because there is no systematic, objective
content - and adds his own twist to the story he receives, before
passing it on. Sometimes the twist is slight, sometimes it can be
critical. Insofar as it substitutes opinion for fact, it's always
undesirable.
6.0.5 Responsiveness
Projects very rarely fail catastrophically, completely without
warning. There is almost always a history, some process or sequence
of developments that leads up to apparently sudden crises. The
problem is that planning systems are simply not designed to capture
the sort of historical information that might be useful in detecting
these sequences or trends.
They capture reasonably well the "instant in time" snapshot
pictures, cross-sections through the history of the project that are
necessary for short term activity planning. But these systems are
very poor at knitting these cross-sections together or in other ways
producing longitudinal time based views of the project's evolution.
Planning systems simply don't cope with trends very well. So project
teams often don't realise they are off target until too late and it
becomes costly and disruptive to institute corrective actions:
"crash" recovery programmes and suchlike.
6.0.6 Learning
The four problems outlined above relate to the difficulty of
using planning systems to predict project cost and schedule
outcomes, while the project is under way. The fifth problem is more
to do with accurately establishing those targets in the first place.
It relates to the way in which firms gather and use actual
performance data from their projects.
The individual people who work on projects learn a great deal
from every job they do. The companies they work for learn almost
nothing. The human learning is, for the most part, experiential and
unstructured. Companies are not able to 'learn' in that sense.
Companies learn by gathering structured data that can be analysed,
evaluated and reused in future activities. In order to capture this
material the data has to be specified systematically and there must
exist organising frameworks or other mechanisms that enable it to be
gathered and arranged efficiently for future use. CPM based planning
systems meet neither of these conditions: the data used are too
subjective and the activity models are too job-specific to be
systematically useful on future projects.
6.0.7 Project Management Summary
The issues above have been elaborated upon because they are
rarely, if ever, drawn out in discussions of project failure -
despite the fact that unless the problems they represent are solved,
other attempts to improve construction project predictability will
almost certainly fail.
To summarise, our theory suggests that projects fail because
conventional project management methods and systems:
- Depend too much on intuitive, subjective definition of work
scope and progress assessment
- Are dangerously top-down in their operation, lacking systematic
connection with the production level in projects
- Are inherently poor for forecasting and for trend detection and
analysis
- Provide no effective frameworks or methods for the capture,
analysis and reuse of performance data.
Similar problems occur in the area of cost control. Again, the
principal problem is that the project scope is not specified
systematically or in useful detail. Cost models based on Cost
Planning techniques - elemental costs per square metre for example -
suffer from the same inherent problem of subjective definition as
activity models, with the same results.
Bills of Quantities (BQ) at least have the merit of being based
on the detail of more or less complete design documentation.
However, as discussed at Section 5 above, in order to make a BQ
manageable as a paper document, most of the useful information
recorded in its development has to be compressed out of the final
product. Hundreds of lines of valuable, specific data about
individual concrete columns for example, may in some cases be
reduced to three or four highly summarised line items in the bill.
This is an extraordinary waste of valuable information as well as
being a source of ongoing dispute amongst the members of the greater
project team.
There are other areas where conventional Planning and Cost
Control techniques fail, for the same general reason. Thus, in the
absence of a well specified, shared baseline work scope, it is
impossible to establish a direct or useful correspondence between
cost and planning systems. So the cost and schedule dimensions of
the project get out of sync, tell different stories, provide
different feed-back to management, generally adding confusion to the
picture.
Also, in the absence of a common specification of the scope of
the project, it is more or less impossible to carry out useful
analyses of the joint cost and schedule impacts of change proposals
or other issues. One could go on...
The point is that any given construction project manager,
starting out on a new project, equipped with all the technologies
and management tools the modern industry has to offer, faces a less
than 50/50 probability of success. The strong likelihood is that,
for all the stress and anxiety that he - and all the other managers
on the job - will endure, he will still fail to deliver the thing on
time or within budget.
So, what can BIM do to improve this situation. Specifically, how
can BIM help teams to deliver their projects more predictability?
6.1 BIM-based Construction Project Management
BIM models will assist in construction in two main ways. The
first of these involves simply using the graphical power of
sophisticated 3D models as a means of visualising how the building
fits together. This includes clash detection and construction
simulation exercises, as well as cutting special views of parts of
the building to assist in particular construction tasks. This is not
strictly using the Information component of BIM models, but is
likely to be the application area that drives the use of BIM onto
the construction site. A more fundamental application of BIM, in the
full sense, lies in its use as a tool for managing production on
construction sites. This section discusses the background issues,
the motivation for its use and the probable mode of operation of BIM
for this purpose.
6.1.1 The Construction Site as a Production Environment.
In an orderly production environment the most reliable way of
forecasting how long a given process has to run, at any given point
in time, is by taking the total amount of material to be produced,
subtracting the amount already produced and dividing the balance by
the rate of production achieved to date. The main issues that affect
the accuracy of the forecast will be the accuracy with which the
variables: total output, output achieved to date, and average rate
of production are known. This is the essence of the production
management concept.
Two observations indicate that it may be valid to apply this
approach to construction. First, although it may not appear obvious,
the buzz of commotion on a typical construction site in fact
represents a great deal of highly ordered, systematic activity. The
work tasks are being carried out by experienced, specialist
operatives employed by a variety of specialist contracting firms.
For each of these individuals and firms the task at hand is just
like every other job they do. The particular configuration of the
building, its location, and the other people working on the project
are different, but the basic specialist operations being carried out
are identical from project to project. In a sense, that's what
specialisation is all about.
Secondly, although again it may not seem obvious, buildings are
in fact highly standardised things - if one considers them at the
level of the individual components that make them up. At the
component level, the construction and fabric of apparently different
buildings are actually very similar. Almost by law, a steel beam in
one building that conforms to BS 4360, is identical to a beam of the
same specification in any other building. Similarly with pipework,
doors, concrete or any other component; if it's described by a
British Standard or an ISO standard, it's essentially the same thing
regardless of the particular building in which it occurs. In this
sense all modern buildings can be considered as being simply massive
aggregations of standard entities. The buildings at the aggregated
level may be unique; the components of which they comprise are
completely standardised.
So, the construction industry is made up of large numbers of
people and firms who, essentially, do the same things, to the same
things, on every project they work on. There may be a lot of
distracting 'noise' in the system - weather, ground conditions,
congestion and such like - but most of the activity on a typical
construction site is actually highly systematic, generating
standardised outputs in a methodical, repetitive fashion. This
implies strongly that the construction site can reasonably be
regarded as being an orderly production environment and that it is
therefore appropriate to use production management techniques to
control operations at this level on projects.
The production management approach overcomes all four of the
problems with project management discussed above:
- It eliminates dependence on intuitive, subjective definition
of work scope and progress assessment.
- It connects management directly to the work face.
- It enables dramatically improved forecasting and trend
monitoring.
- It provides a comprehensive framework within which to gather,
analyse and re-deploy performance data from one project to the next.
The problem with production management is that the sheer volume
of information generated at the production level on construction
projects is enormous. It requires that a record is made every time a
steel fixer bolts up a beam, a carpenter installs a door set, or a
plumber installs a spool of pipe. For every event like this a small
packet of data has to be captured to record the relevant details. In
the case of the steel beam this data might include: the piece number
of the beam, the fixer's team identifier, the location in the
building and the date and time.
Optimism bias and a related behaviour,
strategic misrepresentation, are two problems with estimating
and forecasting, much studied by Bent Flyvjberg. See Bent
Flyvjberg, Nils Bruzelius and Werner Rothengatter.Megaprojects
and Risk: An Anatomy of Ambition.
(Cambridge University Press, 2003)