Sunteți pe pagina 1din 6

2008 AACE INTERNATIONAL TRANSACTIONS

EST.S02
Cost Contingency - Analysis and Evaluation
Ms. Rashmi Prasad

ontingency is defined as the amount to be added to the competitive base estimate (prepared realistically for
the known scope assuming typical site/market conditions/oversights/unknowns and is free of excessive
allowances/markups) as per historical experience so that the final project cost estimate has an equal chance
of falling above/below the actual cost or the initial quantifiable estimate. It is intended to mitigate/eliminate the
adverse impacts of the unforeseen or under-predicted events. It is a reserve and hedge against risk for estimate
assumptions (regarding budget estimates, activity durations and sequencing, resource requirement, historical
benchmarking or database inputs being appropriate/repetitive, resource availability, vendor cooperation and
compliance, responses/inputs from stakeholders and/or third parties, strikes, bankruptcies, calamitous weather) but
unbudgeted or unaccounted for scope growth increase or interest rate changes. It is a function of design definition
(process, utilities, facilities and revamp), estimating method (database and level of detail), estimator experience
(type of project/industry), time frame, schedule probability, new technology or prototype engineering, remoteness of
jobsite, engineering progress at time of estimate and material commitment at the time of estimate. It is not a
function of changes in the defined scope of a project (change in capacity or product slate) or for unforeseeable
circumstances beyond management's control (100-year storms or strikes against equipment vendors). It is money
that is expected to be spent, hence, controlled.
CONTINGENCY ANALYSIS
The concepts for contingency formulation is based primarily on contract type being fixed fee/lump sum (where
contingency is restricted to a minimum) or cost plus (where normal contingency is applied as required for budget
restraints, engineering schedule versus construction schedule, client/project requirements, material availability,
warehousing and laydown facilities and special material control /procurements systems). Some of the commonly
used methods are detailed below:
Predetermined or Mandated Percentage of the Base Estimate
It is an easy and consistent method but unsubjective, inflexible and unspecific to potential risks (as process
complexity, use of new technologies and level of project definition). It produces large variations in the probability of
overrun or under run.
Expert's Educated Judgment
It is an expanded version of the predetermined contingency approach where bounds or norms (formal or
informal) for contingency outcomes are considered. The experts select contingencies predetermined for discrete risk
levels (15 percent for high risk, 10 for average and 5 for low risk) varying with skill, knowledge, experience,
thoroughness and motivation. These are input for all items in the cost summary sheet and linked with a probability
distribution factor. The Monte Carlo run to randomly sample each number produces a population of total estimated
costs from which statistical characteristics of the modeled output is computed.
However, no clear guidance for determination of the minimum and maximum values for each input variable
exists. Unless detailed notes are kept, it becomes difficult to reconstruct the thought process used to establish those
ranges and maintain consistency throughout the process. If a single source of uncertainty affects multiple cost
components, then the full impact on cost risk may not be realized if it is fragmented. Say, if material pricing for
concrete and steel is developed from the same cost data book and these cost components are modeled separately,
then the resulting simulated total cost for concrete and steel may be reduced because of the random cancellation
between sampling of the input variables. The more a single source of uncertainty is fragmented, the less impact it
will have on the total output. Again, the analyst can fall into a trap of focusing on an expected outcome and stop
EST.S02. 1

2008 AACE INTERNATIONAL TRANSACTIONS


thinking about whether the ranges on the cost input make sense making the risk model stop serving its function of
providing answers about the uncertainty in the estimate and instead becoming an exercise in manipulation.
Integrated Schedule Risk Analysis (SRA) and Cost Risk Analysis (CRA)
It involves conducting SRA using risk factors, determining its effect on cost risk (in terms of labor and
management hours and escalation), finding risk factors in CRA and then decomposing the result. The method
focuses on the risks (not the activities or cost line-items which are impacted by the risks), the probability of success
and contingency reserve needed for any degree of safety. It helps to prioritize the risk list and explain management
how the contingency occurred and focus risk responses on the risks, thus, elongating the duration of work for time
dependent resources (labor, supervision), thereby, increasing the cost estimate.
Consider a schedule for construction of a new refinery with risk drivers and hammocks (for CRA).Run SRA to
get probability of finish dates, 80th percentile and P10 => P90. The order of contribution to the time contingency
(the smoothness of progression being dependent on the structure of the schedule) is determined. The risk register is
developed by decomposing the overall schedule risk into its sub-components to help management understand the
rationale of contingency reserve and target risk mitigation as the traditional 3-point estimate for activity durations
tells only which activities are crucial and not which risks are driving. The risk register (already compiled) is used to
identify high-priority risks with their probability through interviews and workshops with knowledgeable people and
impact through a 3-point factor to reflect the uncertainty in the duration of activities it is assigned to (Quantitative
Risk Analysis) followed by risk identification (Qualitative Risk Analysis). The risks are assigned to schedule
activities (a risk can be assigned to several activities and an activity can have several risks assigned), Monte Carlo
simulation computed and results analyzed for estimate sensitivity and net effect of key risks.
The probability and impact data are collected for all identified risks and assigned to schedule activities. The
essence of risk lies in its uncertainty of its occurrence (specified by a probability) and its impact (specified by a
range of durations). If the risk may or may not occur, the probability that it will occur is specified. When the risk
occurs , it affects the activities it is assigned to on X percent of the iterations, chosen at random. On (1 X) percent
of the iterations, the planned value is used. If more than one risk is acting on an activity, the resulting ranges should
be multiplication of the percentages. The risk factors should be selected at random for each risk driver, then the
assigned risk factors should be multiplied together and the resulting factor should be multiplied by the activity
duration for that iteration. Traditionally SRA calculates the sensitivity of the schedule risk for the activities that are
risky and is not satisfactory to emphasize only the effect of the risks and not the risks themselves. However, the risk
driver approach specifically emphasizes which risks to mitigate and not only the risky activity or risky path. It
explains the time contingency to the 80th percentile by establishing the link between risks and its effects explicitly
to determine most cost-effective mitigation strategy along with the risk driver tornado to determine most important
risk. The schedule risk with the highest priority is integrated into the cost estimate and risk driver method is again
used for CRA of baseline estimate plus escalation minus contingency. The SRA results are used to determine the
time-dependent costs (by dividing labor estimate by the months in baseline schedule) as monthly burn rate (by
multiplying the burn rate with the number of months from the schedule risk) and using the results to drive escalation
by escalating the costs to the centroid of its spending activity.
The risk drivers in the CRA are assigned as three-point impact in percentage to the activities they affect and a
factor is selected at random from the 3-point estimate for each iteration. The factor is multiplied with the baseline
cost of the line items to which it is assigned for that iteration. If two or more factors are assigned, they are all
multiplied by the cost. If no factor occurs in an iteration, its value is 1.0 and has no effect on the duration. For timerelated costs like labor, they affect the burn rate while for time-unrelated costs like equipment, they affect the
estimate directly and have a probability of occurring. If they occur, their value is taken at random from their
distribution (usually triangular). If they do not occur on some iteration, their value is 1.0 and they do not affect the
result. Average burn rate do not account for ramp-up and ramp-down or mobilization and de-mobilization. For timerelated costs, the SRA results are substituted for the baseline distributions which are fitted from the simulation
results (3,000 iterations) using crystal ball fit and then correlated in the CRA because it would be unrealistic that one
span would be long and another short in the same iteration. The escalation of contingency is finally calculated like
any other cost by correlating the durations used for the months to the centroid of the activity being escalated or
beyond in CRA.
Influence Factor Based Risk Analysis
It is based on a standard list of interview questions for all possible factors which introduces a source of
uncertainty around all major estimating work activities of the estimate having typical yes or no answers. The way
the questions are answered will either add cost or have no effect. The percent uncertainty from each question is
EST.S02. 2

2008 AACE INTERNATIONAL TRANSACTIONS


multiplied by its associated cost and then summed to arrive at the total cost of uncertainty to identify where
improvements can be made in the work process to increase the estimate accuracy.
The estimate development questions are either generic such as estimating softwares used to generate quantities
and price, documentation supporting the effort or specific such as scope and quantity development (availability of
complete list of equipment, preliminary data sheets, factored bulk quantities, plot plan, electrical one-line diagrams),
material pricing (use in-house quotes, telephone quotes, procurement plan, expedite fabrication), productivity
/schedule considerations, subcontract cost, construction management cost and home office cost. A probability
distribution function is assigned to each question to represent the percent of uncertainty introduced into the estimate.
If the answer to the question of whether the plot plan was available is no, then it introduces a specific uncertainty
range in the cost. As the questions addresses the work process, the percent of uncertainty is standard for all estimates
(regardless of who runs the model), thus, generating the same results with the differences only with the way the
individual answer the questions.
A cost is associated from the estimate to each question in the interview say, bulk material cost is linked with
uncertainty in generation of bulk quantities and labor cost is linked with uncertainty in using unskilled labor. The
cost of uncertainty for each question is the percent from the probability function multiplied by the cost affected by
the source of uncertainty i.e. Cost of Uncertainty = (Percent Uncertainty) x (Estimated Cost). The cost of uncertainty
from all the questions in the interview is summed to arrive at the total cost of uncertainty plus the base estimated
cost.
Time Based Risk Analysis
It is based on defining contingency as minimum (where no risks are fired, no impacts are counted, is always 0
percent), expected (calculated mathematically) and maximum (where all risks are fired, all impacts are summed) and
mapped with the first, last and expected points on a graph showing every risk with level of impact and probability of
firing.
A graph is drawn between contingency applied to duration (on x-axis) and probability (on y-axis) from which
most likely contingency is calculated from impact and probability of all risks. As the x-axis is contingency range, a
function f(x) is applied to points along that range to derive their y-axis component. The impacts from any risks that
are fired are summed to find a total impact for the project by generating a random probability from 0 to 1 and adding
it to the probability of the risk (also 0 to 1). If the combined probabilities >= 1, then the risk is considered to have
fired i.e. risks defined with a larger probability will have a greater chance of being randomly selected and vice-versa.
This generates one possible contingency for the project which may not necessarily be the correct contingency.
The above procedure is repeated to build a sample set of potential contingencies for the project between the
minimum and maximum contingency values. The range of results is divided into groups of equal size which forms
the divisions along the x-axis. Say, a project with range of contingencies from minimum = 0 percent to maximum =
30 percent is divided into six groups each of 5 percent size and number of results falling into each group is counted.
Group 0 - 0% < x < 5% = 0 results
Group 1 - 5% < x < 10% = 200 results
Group 2 - 10% < x < 15% = 400 results
Group 3 - 15% < x < 20% = 500 results
Group 4 - 20% < x < 25% = 600 results
Group 5 - 25% < x < 30% = 250 results
The above data is converted into actual probabilities by dividing the number of results in each group by the total
number of results.
Group 0 => 0 / 2000 = 0.0 probability
Group 1 => 250 / 2000 = 0.125 probability
Group 2 => 400 / 2000 = 0.2 probability
Group 3 => 500 / 2000 = 0.25 probability
Group 4 => 600 / 2000 = 0.3 probability
Group 5 => 250 / 2000 = 0.125 probability
These probabilities should always total to 1 and the project has a probability of 1 of completing in one of these
groups. Again, a graph is plotted with groups (on x-axis) and contingency duration (on y-axis) where every point
EST.S02. 3

2008 AACE INTERNATIONAL TRANSACTIONS


will indicate the probability of any one duration being correct including the cumulative probabilities of all
contingencies up to and including the point in question.
Contingency Ratio (CR) for Process Industry
CR is defined as the contingency amount divided by the subtotal estimate and is correlated with variables
relevant to contingency evaluation such as estimate type/degree of definition, owner company type/experience and
project innovation/feedstock. The accuracy of contingency adopted in actual practice is evaluated by accounting the
difference between the estimates and actual completion costs. A series of regression analysis is performed to arrive
at statistical models which test the dependency of CR as a function of subset of the independent variable(s).
If the variables regress together with estimate combinations, R Squared and F is considered simultaneously
and if the contribution of each of the variable is independent, B value (parameter estimate) is only considered. The
variables include project definition for site-specific data (definitive and completed, preliminary and limited, assumed
or implicit, unavailable), amount of engineering design completed at the time of estimate (engineering completed
and design specifications exist, work to date is characterized by study design of moderate to extensive nature,
limited basis, screening), process type for methods of chemistry , construction and operation (a new design is
necessary for unit at that scale or new materials for construction are required, unit is to be used with operating
conditions that have not been tried with that design and scale, the chemistry to be used has not been demonstrated in
commercial use with that piece of equipment, the combination of feedstock and equipment has not been used tried),
percentage of estimated capital costs incorporating new technology, percentage of heat and material balance
equations (known implying previous experience existence, unknown implying innovation) from commercial
experience at the time of cost estimate, contract type (cost-reimbursable, all other), contingency estimate determiner
(estimator, project engineer, top management).
It helps to quantify well in advance what would otherwise be considered a overrun. The measure of efficiency is
based on how well the contingency amount covers the difference between estimate subtotal and final actual cost by
comparing CR (=Contingency Amount/Estimate Subtotal Amount) to the Overrun Ratio (OR=Overrun
amount/Estimate Subtotal Amount). Ideally, the two should be equal. Realistically, OR is always higher than CR. If
an estimate is woefully in accurate/grossly under or overestimated, then the overrun and contingency curve will
never converge. Even if the paucity of information at early stages preclude accurate estimates, it is expected that as
the information becomes available, the estimates should eventually become accurate. If the contingency estimate
procedures available do not correspond with the owners CR estimation concepts, it requires improvement of basic
subsystem estimating techniques by including innovation and experience of all stakeholders to admit uncertainties
and identify potential significant risks.
Contingency Factor Rate (CFR) for Process Industry
It involves determining the key factors that impacts accuracy of the cost estimate by assigning them a relative
weight to get a Contingency Factor Rate (in percent). A contingency chart is developed from historical data of
several projects with estimating contingency experience with CFR percent(on x axis) and contingency percent (on
y axis). The scoring of factors for contingency will generally include basic process design (for process flow sheet,
heat and material balance, capacities and flow rates, mechanical equipment list, PandIDs), estimators experience and
cost database (for relevant experience of estimators, accuracy and reliability of correct cost database, adequate input
from design and project personnel, feedback, current resource surveys at project location), time allowed to develop
estimate, project and site conditions (schedule and preliminary execution plan, site plot plans and layouts, temporary
facilities and access, site and safety regulations, regulatory design criteria, environmental assessment), current
business and labor conditions (competitive market climate, contracting strategy, future risks on costs and time
baselines, escalation and exchange rates, labor productivity and craft rates) and teams experience and inputs (project
manager involvement, constructability and construction preplanning, third party licensors and suppliers).
CONTINGENCY EVALUATION
The Correct Evaluation Of Contingency Requires Determining the Contingency Type.
Estimate contingency is an allowance for undefined items of work which will have to be performed or elements
of costs which will have to be incurred within the defined scope of work covered by the estimate and which could
not be explicitly foreseen or described at the time the estimate was made because of lack of complete, accurate and
detailed information. It is calculated as part of the TIC estimate through Monte Carlo analysis with 95 percent
certainty of staying within the forecasted amount. It may ultimately be used to fund project deviations or reduced
with passage of time (and risk) for estimating inaccuracies/errors /omissions, quantity growths, project performance.
EST.S02. 4

2008 AACE INTERNATIONAL TRANSACTIONS


Examples are pile count/costs greater than estimated, contract unit rates varying from budget, availability/haul
distance for fill material, soil conditions resulting in need for dewatering and mud mats, footing/foundation depth
changing because of interpretation of soils report, size of foundation changing because of vendor selected, forgot
containment curbed areas and/or lining of containment areas, miscellaneous steel increased for electrical supports,
missing the sump pump, fabrication errors by equipment vendor increasing field costs, pump size increasing to
comply with project requirements, location of cooling tower changing cooling water lines, rework because of
pipe/elect interference, heat tracing quantities changing as scope is better defined by engineering, quality of lighting
panel boards increasing as a result of lighting design, quantity of steam traps increasing per process and piping
design, productivity varying from budget because of work hours, lay-up of equipment after chemical cleaning, per
diem estimate for staff was understated, construction equipment and/or heavy haul costs underestimated, recycle of
engineering documents, coordination issues resulting from execution of engineering effort. Event Driven Risk
Contingencies is the outcome of the BRMF process which is updated at regular intervals (particularly when risk
trigger events have occurred or passed when a new run of the BRMF is run) to determine the contingency that still
needs to be maintained for project execution decisions, engineering errors, difference in contract interpretations or
schedule impact. Examples are soil replacement required to stabilize site, concrete settlement because of using mat
foundations in lieu of piling, mass excavation versus individual foundation excavation, method of support changing
from driven piles to caissons, decision on steam turbine vendor resulting in a down draft turbine requiring a pedestal
foundation and not a block, demolition of foundation that was poured incorrectly, pouring concrete during winter
months, access platforms added for OandM or safety reasons, shop loading delays delivery, selection of steam
turbine vendor results in a down draft turbine requiring a taller building, lifting lugs crack resulting in delays in
installation, transportation damages occurring to equipment in transit, cost/schedule impact because of direct hire or
subcontracting of equipment erection, design for HP steam requiring a more costly alloy piping material, market
price for material increases, main transformer failing electrical tests, motor load increase requiring additional
electrical equipment, DCSs FAT or SAT failure, DCS programming errors, replacing insulation with acoustical
insulation, weather impacts to paint and insulation effort and changes in craft work.
It Should be Differentiated from Budget Allowances and Management Reserves Which Are Different Project
Risk management concepts. Allowances are derived from events which are expected to occur within the scope of the
project and are not risk-based or dependent. Material take off allowance is applied for quantity development as
design completion reaches 100 percent from incomplete engineering and evaluated based on engineer/estimators
assessment of engineering definition, drawings, sketches, work progress and historical performance. Design
development allowance is applied for minor changes made to mechanical equipment and incompletely engineered
items (as electrical equipment, major instruments, additional nozzles, lifting lugs) to equipment pricing after orders
have been placed. Both should be managed by contractor as part of normal project forecast. Reserve funds are used
for changing and evolving scope and deliverables. An owners suspicion is that it is slush fund to offset errors
resulting from poor engineering or estimating as a small, all-encompassing allowance to absorb all unforeseen
project costs. Differences in understanding of contingency can lead to arbitrary reduction (by owner) of the
contingency allowance with unfortunate consequences, such as over-runs and unexpected lower project financial
returns. It is vital the owner clearly understands contingency and estimates risk. Contingency should not include for
scope change for which owner should have a reserve fund.
The Manner in Which Contingency Funding Is Developed Dictates the Guidelines of Its Management for the
Realization of Risks.
Since the manner in which risk affects a project is a function of knowledge, time and status which changes
relative to each other at different points throughout the project, an ad hoc 10 to 15 percent of the total budget for
contingency does not justify the adequacy of the line item contingency for the risks and when the contingency
should be used. It should include a baseline cost estimate with accompanying basis of estimate and risk assessment
to rationally identify the impact of risks on the cost/ schedule of a project. The assumptions for lead times for
procured items, activity sequencing and durations (based on assumed productivities), major cost assumptions, price
quotations from vendors, productivity rates, engineering design options and construction planning should be
understood and regularly reviewed. The risk management should involve steps to identify, analyze, plan, track,
control as a continuous activity and not a one-time activity completed once contingency is determined. It should not
be hold throughout the project to be spent only at the end of the project. It should be sufficient to cover the costs or
time required to avoid, transfer, mitigate or bear the realization of risks with time and completion progress. The
available contingency should reflect these changes by readjustments through periodic reassessment and risk
quantification. It should be large enough but not exceeding needs as then the company is deprived of funds that
might be better use in other ways or on other projects as unspent contingency will find their way to funding scope
changes, enhancements and other elements that should properly be purchased with. For a one-off or one time project
in which it is imperative that the budget should not be exceeded, it should be set a higher point on the distribution
EST.S02. 5

2008 AACE INTERNATIONAL TRANSACTIONS


than the mean value. For multiple projects being executed at any given point in time, all projects should be funded at
the mean whereby some projects will come under the budget and others will fall over the budget, thereby
minimizing the net effect on the total budget for all the projects. It must be depleted either pessimistically (showing
remaining contingency in forecast cost at completion and flagging over-runs when the forecast exceeds base
estimate plus contingency) or optimistically.
The Contract Type Affects the Burden of Contingency Funding Management Where whosoever underwrites
the risk of budget and/or schedule overruns should be capable of applying or utilizing contingency. In lump sum
contracts, the contractor undertakes the project for a stipulated price and any expenditure exceeding that price (save
for reimbursable scope changes) are contractors with no risk for the owner. In time and material or cost plus
contracts, the contractor undertakes the project and is reimbursed for all costs encountered, any expenditure
exceeding the project budget estimate is borne by owner and any under runs are returned to owner which is the sole
party at risk. In guaranteed maximum price contracts, the contractor undertakes the project for a stipulated not to
exceed price. Any expenditures exceeding this price are the contractors. Any under runs are returned to owner who
while not being at risk of an overrun has an opportunity risk of forfeiting potential under runs. In target with
incentives contracts, the contractor undertakes the project for a target price in which both owner and the contractor
participate in overruns or under runs and share risks on an equal basis.

he total project cost contingency versus agreed to probability of success should be determined by
probabilistic ( not algebraic) summation of the individual residual risks. The allocation should be done at the
subsystem entities instead of the task level. The method of calculation should be statistically verifiable based
on mathematically sound axioms and not on use of feel of human judgment. A P50 value with upper and lower
limits as P90 and P10 value is best to keep the estimate value within the bounds 80 percent of the time. Since it is
estimated based on the confidence level in each part of the project, it should be re-evaluated periodically during
implementation and forecasted to reflect actual needs particularly at the end of each project phase. It should not be
distributed or transferred to cover overruns (variances for each package should be measured against the base
estimate) or automatically depleted to balance the overall project forecast. It should be distributed across the
contingent accounts as per the risk model on time based schedule with proper development of drawdown curve
relating contingency with the amount of activities completed. It is essential to take the time and trouble to establish a
projects definition rating before making an estimate to uncover areas of uncertainty by including contingency.
RECOMMENDED READING
1. Adens, Gillian and Gareth Tuckwell, Risk and Contingency Calculations, Tassc Limited.
2. Bent, James A. Evaluating and Calculating Ccontingency, 2001 AACE International Transactions.
3. Burroughs, Scott.E.and Gob Junting, Exploring Techniques for Contingency Setting, 2004 AACE
International Transactions.
4. Coppo, Jerry and Waymon Lofton, Practical Risk Analysis, 2007 AACE International Transactions.
5. Hi, Karen and Wayne, Why Projects Often Fail Even With High Cost-Contingencies.
6. Hulett, David T. Integrated Cost / Schedule Risk Analysis using Risk Drivers, Hulett and Associates.
7. Kamil, Haitham, Contingency Management EPCM Consultant: Owners Representative, 2003 AACE
International Transactions.
8. Kindinger, J.P. Use of Probabilistic Cost and Schedule Analysis Results for Project Budgeting and
Contingency Analysis at Los Alamos.
9. Milanese, James J. Process Industry Contingency Estimation: A study of the Ability to Account for
Unforeseen Costs., Rand Publication Series.
10. Noor. I. and T. Rye, Guidelines For Successful Risk Facilitating and Analysis, 2001 AACE International
Transactions.
11. Noor, Iqbal and Robert Tichacek, Contingency Misuse and Other Risk Management Pitfalls.
12. Prasad, Rashmi. Schedule and Cost Risk Analysis, 2007 AACE International Transactions.
13. Ripley, Peter.W. Contingency! Who Owns and Manages It?, 2004 AACE International Transactions.
Ms. Rashmi Prasad
Fluor Daniel
INDIA
rashmi.prasad@fluor.com
EST.S02. 6

S-ar putea să vă placă și