Investigating the impact of experience curves on the development of Brazil’s presalt cluster

JOSÉ MIRANDA FORMIGLI FILHO, MAURO YUJI HAYASHI and RENATO DA SILVA PINHEIRO, Petrobras; JEAN LE CORRE, ILSON DALRI, JR., ANDRÉ TRUZZI and HENRIQUE SINATURA, Boston Consulting Group

Applying a new conceptual model to presalt wells and subsea systems in the Santos basin allowed the identification of initiatives that will yield major capex savings for Petrobras.

Petrobras, working in partnership with the Boston Consulting Group (BCG), recently conducted a study applying the concept of experience curves to the development of Brazil’s presalt cluster in the Santos basin. Underlying the study was the belief that the magnitude and duration of the presalt development campaign would strongly drive experience-effects gains via continued optimization efforts. The study, thus, had as its key objective the identification of initiatives that could intensify the experience effects for critical items in the construction of oil wells and installation of subsea systems in the presalt, reducing Petrobras’ expected production development capex over the next 20–30 years. This article provides a brief summary of the applied concept, methodology and results achieved.

CONCEPT: EXPERIENCE CURVES

The experience curve concept, which posits that unit costs for a given product or process will decline at a predictable rate as cumulative production volume increases, was developed by BCG founder Bruce Henderson in 1968. In contrast to the well-known concept of a learning curve, which typically represents a passive observation of short-term gains in repetitive processes, experience curves cover longer periods of time and can encompass a large range of factors, from planning and process optimization to scale effects and the implementation of new technologies. In this context, they can be used to direct investments and managerial efforts to the places where they will yield the most impact. While a cost curve converges to an asymptote as volume increases in a linear scale, on a log-log graphic it approaches a straight line; this has been defined as the experience curve, Fig. 1.

Fig. 1. An experience curve depicts the relationship between unit costs and cumulative volume produced. In a linear graph (left), the relationship between unit cost and cumulative volume tends to an asymptote, while, in a log-log scale, this relationship tends to a straight line. In this example, the experience curve has a slope of 80%, indicating that, for each doubling of the cumulative volume, the unit cost falls 20%.

The slope of an experience curve describes the relationship between cost and volume—specifically, the percentage decrease in unit cost for a given percentage increase in cumulative volume. If a product has an 80% experience slope, its unit costs will decline 20% (1 minus the experience slope) every time cumulative production volume doubles. Experience curve slopes typically range from 70% to 90%, and their calculations in real-world processes involve a complex set of analyses and data adjustments to isolate intrinsic experience effects from those related to commodity prices, exchange rates and inflation, for example.

ESTIMATING EXPERIENCE CURVES FOR PRESALT

The methodology applied in this study involved five main steps, developed by Petrobras’ key managers and technicians and BCG’s upstream experts:
1. Definition of standard scenarios for wells and subsea systems and their respective cost structures
2. Prioritization of critical items according to cost materiality and experience potential
3. Identification of applicable analogs for each critical item and estimation of respective historical experience curve slopes
4. Projection of the experience curves for the prioritized items in the presalt context for the next 20–30 years
5. Consolidation of the individual experience curves for each critical item into single experience curves for a presalt oil well and subsea system.

The first step was to define standard types of wells and subsea system scenarios that would be applicable to the presalt campaign. This analysis led to the following:
• Three types of wells: vertical, vertical lean (an open well without intelligent completion and minor reservoir-evaluation intensity), and open horizontal
• Four subsea collection systems, having as variables the number of interconnected wells, whether or not the systems use manifolds, and whether the systems’ risers are rigid or flexible
• Three subsea gas-export systems based on riser type (auto-sustainable hybrid, rigid or flexible).

For each of these wells, subsea collection and subsea gas-export scenarios, a detailed cost breakdown was developed leading to the prioritization of 10 critical cost items for wells and five critical cost items for subsea systems. Those items included drilling systems (bottomhole assembly, bits, fluids, etc.), reservoir evaluations, rigs and their respective maritime logistics support, subsea trees and manifolds, and the installation of flowlines and umbilicals.

It is important to note that experience curve projections were run initially for each of the selected critical items rather than for the wells and subsea systems of the presalt as a whole. This is because the curve calculation depends on a historical cost base that is technically comparable to the presalt situation. This historical base is not available for presalt wells due to the recent discovery of the Santos basin fields and the limited existence of wells in comparable situations around the world. It is, however, possible to analyze a presalt well through its subcomponents, by identifying applicable analogs for which there might exist historical databases and adjusting these using pertinent normalizations (e.g., adjustments by water depth, well length and geometry, etc.) to calculate the respective individual experience curve for each subcomponent, Table 1. These curves could then be compounded to generate the synthetic experience curve for a whole oil well or subsea system.

Table 1. Examples of analogs to the presalt environment (click to enlarge)

Based on the defined analogs and collected data, the next step was to determine historical slopes for each critical cost item. This entailed the definition and vetting of a number of parameters, such as the ones illustrated below that were used to define the historical experience curve slopes associated with productive rig time in the drilling stage:

Choice of analog. Given limited historical data for presalt drilling, the performance from Petrobras’ postsalt fields was used as an analog. This analog was considered applicable to the presalt due to the finding that, via several simulations, the experience curve related to drilling performance tended to present the same slope, regardless of the lithology, bit type or well geometry in question.

Applied assumptions. Due to the absence of a global database on drilling efficiency, Petrobras’ internal data was used as an approximation of performance for the industry as a whole. Considering that the companies supplying drilling systems are major corporations operating globally, this approach was assumed to faithfully reflect the historical rate of gains in drilling efficiency for the industry.

Experience curve metrics. For the y-axis, hours per meter drilled was used, being a physical measure of drilling efficiency, including useful time and maneuvers. For the x-axis, cumulative drilled meters in the industry was used.

Applied data. Records from Petrobras’ wells were mined for drilling times as the source for the y-axis, and a selection of international databases were used to estimate the reference base as a source of the x-axis.

Required normalizations. Because in this example there were no monetary measures involved (only physical ones; i.e., meters drilled), adjustments to eliminate the effects of inflation or changes in commodity prices were not needed.

Historical curve parameters. To estimate the historical curve parameters, several databases were tested, including a consolidated base of Petrobras’ offshore wells (about 1,200), a base of drilling performances with PDC and tricone bits (65 wells in Albacora field and 69 wells in Barracuda field, respectively), and a base of drilling performances in horizontal and vertical wells (about 40 wells in Albacora).
Estimation of slopes for each of the samples was performed through a regression on a logarithmic base (log-log) according to the formula:

T_d = a + b · log D

where T_d is the drilling time per meter and D is the drilled depth in meters, which yields a slope of 2b. For all of the considered historical data samples (offshore wells in general, verticals and horizontals, with tricone and PDC bits and for different lithologies, such as carbonates), performance gains in drilling across the years presented very similar slopes. Accordingly, the same obtained slope value could be applied to the drilling phases in a presalt well, regardless of the respective lithology of each phase. Figure 2 illustrates sanitized log-log regressions for several considered samples.

Fig. 2. Sanitized examples of log-log regressions for drilling under specific conditions.

This process for estimating historical slopes was performed for each critical item in the wells and subsea systems, with each estimation considering a careful selection of reference bases and applicable analogs, as illustrated in the example above. Having obtained those historical slopes, it was now possible to project how the costs of each critical item would evolve for the next 20–30 years given the experience effects over that time period. To do so, three key parameters had to be estimated:

Current cost reference for each item in the presalt context (C₀). This is the item’s unit cost today, from which experience effects are applied at each doubling in the cumulative production volume. It is important to emphasize that the initial cost C₀ should be the one from which the respective production process leaves the “laboratorial” stage (during which the unit costs exhibit erratic behavior, with no statistical meaning), migrating to a more standardized production process (from which experience curve effects start to materialize). Another important consideration is the definition of a proper metric for the unit cost measurement. For instance, for the experience curve for drilling bits, several metrics could be used, such as cost per bit, well, phase or drilled meter.

Volume reference base (V₀). This is the current cumulative volume from which volume duplications (doublings) will be measured. To assess this parameter, it is vital to be clear about the reference base applicable to the item under consideration. In the case of subsea trees, the reference base applicable to the presalt is that of subsea trees in deepwater and ultra-deepwater conditions, given similar operating environment characteristics. In contrast, subsea trees used in shallow water have very distinctive characteristics, including different installation processes, and, therefore, could not be used as a reference for the presalt. Thus, when considering volume projections for presalt subsea trees, the starting reference base should be the total volume of all deepwater and ultra-deepwater subsea trees installed up to the current time.

Slope. This would be the historical slope estimated as described previously in this paper using analogs for which historical cost bases and operational performance indicators were available.

Once these three parameters (C₀, V₀ and slope) were obtained for each critical item applied in presalt well construction and subsea systems, the respective projections of cost (or performance) for each item would then be directly dependent on the future duplications expectation of their respective volume reference bases. Projections could thus be made for how all of the relevant reference bases would evolve for the next 20 years and for the expected cost reduction driven by experience effects.

After completing projections for individual experience curves for each item, consolidated experience curves for an oil well and for a subsea system as a whole were estimated. That estimation was done through a complex simulation model developed by BCG, which compatibilized the different reference bases (V₀) of the individual items’ experience curves into one single reference base (number of wells) allowing the projection of consolidated cost curves for different types of wells and subsea systems considered for the presalt. As inputs, this model receives the individual experience curves estimated for the critical items and some other specific characteristics of the presalt campaign, including water depth, well length and rock formation composition.

Based on the estimation of consolidated experience curves, the potential for investment reduction in wells and subsea systems across the presalt campaign could be calculated. In present value, these investment reductions added up to 11% for well drilling and completion and 10% for subsea systems. Combined, these savings amounted to about 8% of the total forecast investment for the presalt campaign (including FPSOs).

INTENSIFYING EXPERIENCE EFFECTS

The approach used to seek opportunities to intensify experience effects in the presalt included analysis of key experience levers (identified by running simulations in the model described above) and a series of workshops with Petrobras expert teams by function and key processes. During these discussions, the technical teams identified typical experience-effect intensification patterns, reflected in increased slope or downward vertical shift of the curve, Fig. 3. As a result of these exercises, more than 150 intensification initiatives were identified, with 30 of them being prioritized as having the greatest potential impact for the presalt wells and subsea systems. The identified initiatives are diverse in nature and include not only technical matters related to the concept, planning and execution of the offshore production systems, but also organizational considerations, performance management drivers and supplier relationship development. Three of these prioritized initiatives are illustrated below:

Fig. 3. Experience curve intensification patterns.

Reservoir evaluation prioritization. Given the heterogeneity of the presalt reservoirs and the magnitude of the development campaign, the value of the information from well testing and logging tends to be quite significant. On the other hand, these reservoir evaluations can be very costly to perform in the presalt context. The experience curve approach recognizes the fact that, as these evaluations are performed, Petrobras will gradually reduce the uncertainty (or gain experience) on how to best exploit the fields, changing its reservoir evaluation needs from more sophisticated methods (drillstem tests conducted by rigs and complete sets of wireline logging) to simpler mixes of evaluations (production tests with or without bottomhole closure by intelligent completion systems and simpler logging sets).

From this basic concept, a detailed initiative to identify the optimum mix of reservoir evaluations was conducted for all production modules and blocks in the presalt, shedding light not only on how information through time would benefit the individual well, but also on how the evaluation of each specific well impacted the reduction of uncertainty of its module and field. As a result, this initiative yielded a 33% cost reduction in the presalt planned investments in reservoir evaluations, while ensuring the same value from the information gathered in such evaluations as in the original plan. 

Rig specialization model. The expected scale of the production development campaign in the presalt allowed the analysis of a new rig allocation model, in which specialized rigs, with shorter operating cycles, would be deployed for the construction of specific parts of the well instead of having only one type of rig responsible for all drilling and completion activities. This model would allow not only the intensification of experience effects through specialization and faster ramp-up of crew and equipment performances, but also an optimized use of the asset base, allocating the most sophisticated rigs only to the most demanding steps of the well construction process. In this initiative, two large stochastic models were developed. The first model tested the economic feasibility of different specialization scenarios (which, at the limit, included up to eight different types of vessels/rigs participating in the construction of a well) and generated Petrobras’ demand curve for each type of rig for the next 20 years (considering a 95% service level to meet the desired exploitation plan and production curve). The second model then used this specialized rig fleet as an input and applied a predefined day-to-day allocation rule for each rig in order to test stochastically the expected degree of fleet idleness.

As a result, a robust optimized specialization scenario has been identified, considering the use of a top-hole driller for the first and second drilling phases (the ones above the salt layer), a sixth-generation rig to perform the more complex drilling (salt and reservoir) and completion activities, and a light workover rig to perform the well testing and subsea tree installation, besides eventual necessary workovers. This initiative has identified savings of at least 9% in the rigs to be allocated in the production development of the presalt fields, something made possible given the concentrated scale of Petrobras E&P operations.

Manifold usage optimization. Traditionally, manifolds present very limited experience effects over time, mainly due to their nonstandardized, project-by-project production characteristics. In this context, an initiative was conducted to analyze the tradeoffs and potential experience effects associated with a more standardized use of manifolds in the sizeable presalt campaign. The initiative analyzed different scenarios and was able to identify gains from the acceleration of experience effects representing roughly 17% of the cost of the overall subsea systems to be deployed in the presalt campaign.

All in all, for each of the 30 prioritized initiatives, the potential for additional cost savings from experience effect intensification was mapped, resulting in an estimated aggregate capex savings of 17% in present value for the presalt campaign as a whole. It is important to note that, although this figure assumes a constant investment level for the FPSOs (as they were not part of the scope of this project), it is certain that experience effects are also applicable to these production units. If these effects were included and correctly estimated, the total aggregate capex savings potential for the presalt campaign would most likely be greater than 20%. A separate internal Petrobras program has been carried out to optimize the investments associated with FPSOs.

Finally, all prioritized initiatives were structured in individual projects, with approaches, work plans and teams assigned for their execution. Those initiatives were then consolidated into an implementation program under the name Presalt Capex Optimization Program (PROINV). In its current stage, this program is being conducted until early 2012 under the coordination of Petrobras’ Master Development Plan for the Santos Basin Presalt Cluster (PLANSAL), with the continued supervision of Petrobras E&P top management and the support of BCG.

CONCLUSIONS

The opportunities identified by applying the experience curve concept to the presalt development campaign are now being rolled out broadly at Petrobras, encompassing not only the presalt projects, but all of its major E&P development projects. In this context, Petrobras has incorporated the experience curve program structure as part of a systemic process to reduce the total cost of wells and subsea systems, aiming for continuous investment optimization over the decades to come. 

ACKNOWLEDGMENT
This article updates IBP 2265-10 presented at the Rio Oil & Gas Expo and Conference held in Rio de Janeiro, Sept. 13–16, 2010.