Skip to main content.

Gas-fired power station cost and schedule uncertainty


This case study outlines a quantitative analysis of the contingency required to address cost and schedule uncertainty associated with building a new gas-fired power station. The project was well into the delivery phase, when attention had shifted from planning and forecasting the work to day-to-day execution. The analysis was a means of reviewing whether the established cost and schedule targets remained realistic. It drew the team’s attention away from immediate pressures, to help them identify areas where additional effort and attention were needed to ensure longer-term success.

Among other things, the case demonstrates the importance of addressing schedule uncertainty when estimating the capital costs of a project.



Our client was building a new gas-fired power station on a greenfield site. The client would also operate the power station. Much of the construction had already been completed.

A primary function of the power station was to service a power supply agreement with a large industrial customer. The capital and operating costs would be important in final negotiations about the power tariff, as well as determining the profitability of the plant for our client.

Commercial arrangements had been agreed with the customer and preliminary contracts were in place. The customer was expanding its plant on the basis of power delivery commitments from our client, who would be liable for penalties if power were not available on the agreed date.

Key milestones were very important commercially to both parties.

Contracts for the supply of natural gas were in place.

The project

The project involved:

  • Construction of the gas pipeline and compressor station
  • Construction of the power station, excluding the combined cycle gas turbine (CCGT) package
  • Supply, installation and commissioning of the CCGT package, to be delivered by a major international supplier under a turnkey contract
  • Construction of the switchyard and high voltage transmission line
  • Commissioning of the completed power station
  • Acquisition of land and rights-of-way for the power station, the gas pipeline and compressor station, and the switchyard and transmission line.

We were asked to help the project team quantify the uncertainty around the residual capital cost.


Modelling uncertainty

An Excel workbook with the simulation add-in @RISK was used to model uncertainty in the schedule and the capital cost estimate. The model included:

  • Uncertainty in the main components of the capital cost that were not related to the project schedule
  • Uncertainty in the project schedule
  • Time-variable components of cost, primarily owner’s costs that are incurred for every month the project lasts, linked to uncertainty in the schedule
  • Penalties and replacement power costs that could be incurred to meet supply commitments if start-up were delayed, which are also linked to schedule uncertainty.

The client’s initial motivation for the analysis was to assess the adequacy of remaining contingency funds in the capital budget. The base cost estimate values and actual expenditure to date were used in the workshops as an aid to understanding the relative significance of the cost components and how they might affect the contingency amount, but they were not modelled explicitly.

The model structure reflected the influence of schedule uncertainty on a number of costs in the estimate (Figure 1).

Figure 1: Model structure

Estimating ranges of uncertainty

Estimating workshops were held with the project team and its specialist advisers to derive uncertainty range values for the model. Uncertainty ranges were developed using the template in Table 1. The elements in this template, and their sequence, are designed to reduce estimating biases where possible.

All the participants were experienced in power station projects like this. Project risk registers, the base cost estimate values and other project information were available in the workshops. The team’s experience and project records all formed part of the information incorporated into the range analysis.

Estimates were updated subsequently for a small number of items as further information became available.

Table 1: Range estimating template

Interpreting the estimates

The P10, most likely and P90 values from each range estimate were interpreted as defining a triangular distribution in the form shown in Figure 2. (For the technically minded, we used the @RISK function RiskTrigen.) This assumes each range estimate has defined an 80-percentile confidence range, and that values below the optimistic assessment can arise one time in ten, as can values above the pessimistic assessment. This is a relatively conservative modelling assumption, in the sense that it allows for a significant chance of values falling outside the assessed range.

Figure 2: Triangular distribution

Analysis outcomes

Contingency cost

The distribution of the total cost less the base cost estimate for the remaining construction is shown as a solid line in Figure 3, where the values have been normalised to preserve the client’s confidentiality. This shows how much funding is required over and above the initial estimate to provide various levels of confidence in having sufficient capital to complete the work. The base contingency amount, established at the start of the project, is shown on the X-axis.

Figure 3 shows on the vertical axis the chance that any particular contingency amount on the horizontal axis will be required. There is a 70% chance that the base contingency will be exceeded. The effect of the late completion penalties can be seen in the long tail on the distribution at the lower percentiles.

Figure 3 also shows the effect of removing schedule uncertainty. Compared to the total contingency, the mean required contingency is reduced (the curve is shifted left), and so is the spread (the curve is steeper). The gap between the curves with and without the effect of the schedule increases from left to right.

Figure 3: Distribution of contingency

Sensitivity analysis of the contingency

Figure 4 shows the most significant contributors to the spread of the contingency distribution, confirming the aggregate outcome shown in Figure 3. Uncertainty in the project schedule is by far the most important factor, as it is the main source of uncertainty associated with:

  • The need to pay for replacement power if start-up is delayed
  • Direct penalties for late completion
  • Construction management and other owner’s costs, most of which are linked directly to the schedule duration.

Uncertainty in the commercial and technical costs of the power station balance of plant are also important. These costs exclude the CCGT package, to be delivered under a turnkey contract that has little cost uncertainty.

Figure 4: Drivers of variation in the total contingency

The relationship between schedule uncertainty and total contingency is illustrated in Figure 5, in which the individual values of schedule-related cost are plotted against the total contingency cost over a large number of simulation iterations. The overall importance of the schedule is indicated by the strong pattern across much of the range, with a stronger relationship at higher values where the pattern narrows; this region is where long schedule delays introduce additional contractual penalties as well as the cost of replacement power to meet supply commitments.

Figure 5: Schedule contribution to the total contingency

Figure 6 shows the drivers of the variation in the contingency if the effects of schedule uncertainty are excluded. The relative importance of apparently low-cost items like commissioning fuel (which would be purchased in the spot market) reflects the advanced stage of completion of the project, when many of the costs for high-value work packages had been locked in contractually so there was little scope for major variations.

Figure 6: Contingency drivers excluding the schedule

As an example, the CCGT package was the largest single cost item, but its cost uncertainty was very low and it does not appear as a driver in Figure 6 at this stage of the project. The design was complete, the manufacture of major equipment was on schedule, and uncertainty was restricted to relatively minor matters like interest charges on late payments, reimbursements of visa costs for overseas technical personnel, additional costs associated with commissioning delays caused by third parties, and potential claims that might arise at contract close.

The HV transmission line was also a turnkey package. It appears in Figure 6 because the final details of the route had not been agreed, which could change the requirements for transmission tower foundations and open repricing options for the supplier, with associated uncertainty.


Timing of the analysis

With the project well into the delivery phase, day-to-day implementation concerns were front of mind for most of the project team. The analysis described in this case study was valuable for the team; the structured discussion in which they participated, using templates based on Table 1, helped to identify areas that had been overlooked, where initial assumptions had been proven wrong or where additional effort might be worthwhile.

This helped the team focus their attention on outstanding matters that were important for project delivery, particularly matters that might reduce the schedule uncertainty.

For example, it highlighted those items that were necessary for the satisfactory completion of third-party turnkey and fixed-price contracts, and without which there might be delays or additional costs, that depended on actions or information to be provided by the project team. Some of these requirements were embedded in the assumptions recorded in the templates, but ensuring the continuing validity of many of them had been given less emphasis than more immediate project delivery pressures.

Schedule uncertainty

Schedule uncertainty is an important driver of cost uncertainty in most projects. While schedule uncertainty must be taken into account in the analysis, a fully-integrated cost and schedule model is rarely needed, as the effect of the schedule on major overheads and other time-dependent costs such as penalties is strongly associated with the overall duration of the work rather than the granular relationships between time and cost at an activity or task level.

In this case the schedule analysis was conducted at a quite high level. It was sufficient for the purpose and a cost-effective use of the team’s effort in the analysis and the workshops. Pictures like Figure 3 and Figure 4 illustrated clearly the importance of the schedule, the value of reducing schedule uncertainty as far as this is possible, the need to make appropriate contractual arrangements, and the need for adequate financial provision for the uncertainty that cannot be eliminated. As is generally the case, additional detail and effort would have absorbed an appreciable amount of the team’s time and would not have provided additional value.

Energy provider
Oil and gas
Services included:
Project risk management
Quantitative modelling
Cost uncertainty