Cost Contingency and Cost Evolvement of Construction Projects in the Preconstruction Phase

Abstract

The current literature discusses the methods to estimate the costs and cost contingency. The literature also distinguishes "known unknowns" and "unknown unknowns" contingencies. Little is written, however, about the evolvement of total project cost estimates during the preconstruction phase of construction projects. Moreover, not many studies are investigating the "known unknowns" and "unknown unknowns" contingencies in real construction projects. Practice expressed the need for getting more insight into the development of the estimated costs of the projects in the preconstruction phase. This paper, therefore, discusses the estimate of the total project costs (and cost contingency) in the preconstruction phases of 29 Dutch flood defense projects using a case study approach. Altogether, the projects have experienced an 11.51% increase in the estimated costs compared to the initial estimates, which is low compared to previous studies. This increase in the cost estimates of the flood defense projects can be explained by "technical" reasons. The investigation of "known unknowns" and "unknown unknowns" contingencies shows that the percentage of the "unknown unknowns" contingency has increased in the preconstruction phase while a reduction was expected. This increase suggests that the projects were not confident about their estimates and the increase can be explained by a lack of experience, organizations' culture, or the phenomenon of "pessimistic bias." Practitioners can avoid "pessimistic bias" behavior by asking for opinions about their estimates and using historical project data. Further research is suggested into realized cost contingency after project execution.

Risk induced contingency cost modeling for power plant projects

“…-planned value of each activity was considered as equal to the completion cost of the similar activity of the previous projects -the probability distribution of each activity was given as an input of the MCS to compute the total project cost -contingency cost of a whole project was predicted as the difference between total cost and the planned cost -contingency for each activity was predicted as the percent contribution of an activity to the project cost variance -contingency and cost performance of each activity can be monitored -cost performance after the end of a reporting period gives an early warning about the project budget and indicates which activity is critical to manage -expert judgment-based prediction can accommodate in the absence of a historical record of previous similar activities/projects -does not require huge data collection and tedious calculation -can handle project uncertainties in a complex environment -subjective cost information can be varied, and prediction accuracy fully depends on the level of skill, knowledge, and experience of the estimator -the total contingency was distributed based on percent weight of an activity cost to the total project cost, without considering the level of risk associated with that activity -activity costs are assumed to be independent and normally distributed Jung et al [30] Integrated risk assessment and contingency cost -identify and analyze cost overrun risks -three variables (O, D, S) instead of two variables (O, S) for risk analysis -improved accuracy of risk assessment and performed better than the traditional lumpsum ration approach to contingency prediction -not a dynamic approach, as no guideline was provided to revise the Reference Methods/models/tools Characteristics Advantages Disadvantages/limitations -the cost variable for a group of risks was calculated as the percent (i.e., the level of risk) of the total cost Maronati and Petrovic [9] MCS -evaluate uncertainties and risks -project costs were predicted considering both correlated and uncorrelated variables -individual cost variables (cost of work, price of materials and equipment) were simulated separately and then combined for modelling project total cost -distribution-free rank correlation among the cost variables was used -MCS handles uncertainties in cost simulation -correlation between the cost variables are taken into consideration to estimate cost uncertainties -the triangular distribution function of a cost item is used instead of crisp value, which facilitates flexibility in decision making regarding budget prediction -experts feel more comfortable to provide 3point costs instead of a single mean and standard deviation of a cost item -expert judgment was used instead of economic parameters (inflation, GDP, etc.) for adjusting those cost items to reflect current market price and supply chain -the model underestimates cost uncertainty or contingency cost compared to the real cost overrun -risk/uncertainty was not delicately assessed and addressed into the contingency cost calculation Hoseini et al [98] Basic statistical analysis (goodness-offit test, mean, standard deviation, distribution pattern, etc.) of the predicted contingency cost in practice based on historical data -Gamma, Beta, and lognormal distributions were presented, and lognormal distribution was found well fitted to the percentage of a contingency cost estimate -trends of contingency cost estimates were studied through preconstruction phases (initiation, project development, and tender and award) -separate contingency estimates are provided for the known-unknown and unknownunknown risks -risk induced contingency cost estimates and follows up the trends of prediction development in different construction phases prior to the execution phase -application of this approach depends on the data recorded from different...…”

Budget and Cost Contingency Cart Models for Power Plant Projects

“…The most frequently available models for predicting PC and CC are basic statistical/deterministic models (Hoseini et al, 2020), probabilistic models (Touran, 2003;Uzzafer, 2013), MCS (Barraza et al, 2007;Chang & Ko, 2017;Hammad et al, 2016;Maronati & Petrovic, 2019;Shahtaheri et al, 2016), fuzzy set theory (Jung et al, 2016;Salah & Moselhi, 2015), fuzzy expert systems (Idrus et al, 2011), fuzzy-Bayesian belief network (Islam et al, 2019), regression models (Thal et al, 2010), and artificial neural networks (ANN) (Diab et al, 2017;Lhee et al, 2012) and machine learning (Bilal & Oyedele, 2020;Chakraborty et al, 2020). A brief analysis of these models is presented below to highlight the importance of the CART model used in this study.…”

“…A brief analysis of these models is presented below to highlight the importance of the CART model used in this study. Hoseini et al (2020) use basic statistical analysis (goodness-of-fit test, mean, standard deviation, distribution pattern, etc.) for establishing a CC in practice based on historical data.…”

Assessment of preconstruction factors in sustainable project management performance

“…One major challenge is delivering projects on time and within the budgeted cost, and previous studies have contributed to defining the concept of construction delay. Practice in construction has expressed the need for further insight into the formation of project estimates during the preconstruction process (Hoseini et al, 2020). Similarly, a well-developed preconstruction plan helps to ensure successful construction projects (Abdelaty et al, 2020).…”