A data-driven approach to identify net-pay cutoffs

PAUL F. WORTHINGTON, Gaffney, Cline & Associates

The proposed approach takes account of rock type and reservoir depletion mechanisms and honors scale of measurement. The outcome includes more exact petrophysical interpretations of hydrocarbon-bearing intervals and more meaningful reservoir models.

Core sampling and laboratory analysis images courtesy of Baker Hughes.

The concepts of net reservoir (good-quality reservoir rock) and net pay (good-quality hydrocarbon-bearing reservoir rock) are intrinsically related. A knowledge of net pay is important for the volumetric estimation of hydrocarbon resources. Yet there is no universal definition of net pay, no general acceptance of its role in integrated reservoir studies, and no recognized method for evaluating it, and there are disparate views on how to make use of it.1 Partly for these reasons, net pay constitutes a major source of uncertainty in volumetric reserves estimates, second only to gross rock volume. This article examines the state of the industry with regard to quantifying net pay using determined formation properties and promotes an improved methodology that avoids drawbacks of traditional approaches. A specific aim is to reduce subjectivity, so that different subsurface teams can deliver more closely aligned net-pay estimates. These matters are especially important during the early stages of field life, when uncertainty in estimated petroleum resources is greatest.

WHAT IS NET PAY?

Net pay is a thickness with units of length, and it can only be measured at a well. It is a subinterval within the gross thickness, comprising those net-reservoir intervals that contain a significant volume of potentially exploitable hydrocarbons. Thus, net pay is a subset of net reservoir. Net-pay subintervals are often aggregated to give a total net pay and thence, by ratio to gross thickness, net-to-gross pay. The quantification of net-to-gross pay lies on the critical path to the estimation of ultimate recovery through the “static volumetric” method. However, for geological mapping and engineering purposes, an aggregated net-to-gross pay is not useful in itself. An inventory needs to be kept of exactly where the pay subintervals are located.

In essence, the net-pay concept leads to the identification at a well of those sections of a reservoir that will produce exploitable hydrocarbons, excluding the rest. Thus, net pay allows recovery efficiency to be evaluated meaningfully against an initial hydrocarbon volume that is contained within the reservoir rock. In other words, recovery efficiency is measured against in-place hydrocarbon volumes in rock that will allow reservoir fluids to be stored and to flow. Otherwise, estimates of recovery efficiency can be distorted by the inclusion of non-contributing volumes (in non-reservoir rock) that will not be produced.

Once net reservoir and net pay have been identified, petrophysical algorithms can be established over these intervals as appropriate. This means that interpretative equations can be founded exclusively on calibrating data from those very same intervals to which they are to be applied. Otherwise, data from non-reservoir intervals might influence the establishment of interpretative algorithms and thereby degrade their application over net reservoir and net pay. Moreover, if there is no separation of net reservoir and net pay from the gross thickness, it will be necessary to characterize the non-reservoir rock to the same degree as the reservoir rock. Given that core analysis of lower-quality rock is more expensive than of reservoir rock, it is a challenge to envisage this happening as part of the evaluation of conventional reservoirs in any cost-conscious culture.

The identification of net pay and the associated elimination of non-reservoir rock form the basis for a more meaningful initialization of a reservoir simulator. It is important to note that the traditional practice of defining net pay using static cutoffs, which demarcate intervals with sufficient hydrocarbon-filled porosity, has been refined here through the use of dynamically conditioned cutoffs, which demarcate intervals with potentially exploitable hydrocarbon-filled porosity. This refinement is in accord with the contemporary approach to net-pay quantification within the context of integrated reservoir studies.² Although there is a point of view that simulation is a catch-all that can account for non-reservoir rock as well as reservoir rock, recent experience has confirmed that if net pay is systematically quantified, the performance of dynamic reservoir models is demonstrably improved in terms of more readily attainable history matches.

HOW IS NET PAY QUANTIFIED?

The earlier literature alluded to the “picking” of net pay according to how it was to be used, so that the intended method of application influenced how net pay was identified. Today, this exercise is largely automated, with the possible exception of single-well completions where decisions have to be made on the spot. Net pay is quantified through the use of petrophysical cutoffs that are applied to well logs. Cutoffs are limiting values of formation parameters that remove non-contributing intervals. The role and application of cutoffs in integrated reservoir studies have been discussed previously.^1,3 Traditionally, a shale volume fraction (V_sh) cutoff is used to identify net sand. A porosity (φ) cutoff is then applied to net sand to delineate net reservoir. Finally, a water saturation (S_w) cutoff is applied to net reservoir to define net pay. Thus, net pay is nested within net reservoir, which, in turn, is nested within net sand.^1,4 Because cutoffs remove non-contributing intervals, they should have a dynamic significance.

The application of these principles (of flow criteria, etc.) in quantifying net reservoir and, thence, net pay calls for an examination of porosity and permeability (k) as represented within a conventional core dataset. This should be done in the light of the recovery mechanism and with appropriate data partitioning (e.g., on the basis of facies type) and honoring of scale (e.g., from core to log).¹ Central to this process is the concept of a “reference parameter,” which allows identification of the limit to flow for a particular reservoir unit or subunit and for a given depletion mechanism. Parameters that can be quantified through downhole measurements are tied back to the reference parameter so that a reference-parameter cutoff can be related to cutoffs for properties that can be determined from well log analysis. This process is handled synergically; i.e., all log-applicable cutoffs are tied back directly or indirectly to the same reference parameter so that all cutoffs have a hydraulic significance. This process is shown schematically in Fig. 1, where the reference parameter is either reservoir quality index, expressed as (k/φ)^0.5, for primary depletion,⁵ or extrapolated end-point relative permeability to oil in the presence of irreducible water, k_ro(S_wirr), for waterflood. A worked example for primary depletion is shown in Fig. 2. If the available core data include special core analysis, there are other approaches that can be adopted. For example, if Dean-Stark extracted water saturations are available, it is possible to groundtruth the use of composite cutoff parameters such as bulk volume water, the product of porosity and water saturation.

Fig. 1. Schematic process for data-driven identification of dynamically conditioned cutoffs: correspondence of reference and conventional parametric cutoffs for a) primary depletion and b) waterflood depletion;3 c) synergic quantification of conventional cutoffs using the reference cutoffs from Fig. 1a and Fig. 1b.1

Fig. 2. Worked example of establishing dynamically conditioned, synergic cutoffs using the methodology of Fig. 1a and Fig. 1c. The reservoir is an oil-bearing sandstone under primary depletion. The porosity cutoff of 0.075 corresponds to a fractional shale volume cutoff of 0.41 and a water saturation cutoff of 0.67.

HOW IS NET PAY USED?

Historically, the main reason for determining net pay has been to obtain a value of net-to-gross pay for the calculation of hydrocarbons in place. It has long been recognized that the distribution of resources in a reservoir is better understood if net pay is analyzed for each depositional unit in turn. This is not equivalent to saying that net pay should be quantified by using rock-type-specific cutoffs, even though this may be required. It is more directed at how to interpolate net pay at the field scale after it has been quantified at discrete wells. Net pay appears within a net-to-gross pay term in the following equation for estimating oil in place and, thence, ultimate economic recovery under primary depletion: 

   Eq. 1

where EUR is the estimated ultimate recovery; GRV  is the gross rock volume; N/G₁ is the net-to-gross pay fraction; φ₁ is the average porosity over the net-pay interval(s); Sh₁ is the average porosity-weighted hydrocarbon saturation Sh over the net-pay interval(s); B is the formation volume factor; and RF is the recovery factor (fraction) to the economic limit.

The use of this equation is rooted within the culture of geological unit correlation and layer averages. It is most commonly applied early in the life of a field, although it does persist in reserve audits and related quick-look evaluations.

The advent of 3D geocellular models has led to a different culture. Grid cells are populated with net-to-gross reservoir and porosity data. The precise approach depends on the coarseness of the grid. For example, net-to-gross reservoir might only be allowed values of zero or unity for fine grids. On the other hand, porosity will usually have to be averaged to obtain the single value to be assigned to a cell that is intersected by a wellbore. Hydrocarbon saturation is assigned through a saturation-height function that, ideally, has been established at the vertical grid-cell scale using net-reservoir inputs. The saturation-height function takes account of the variation of  Sh not just with the properties of a reservoir rock but also with its height above the equilibrium pressure surface associated with the base of the hydrocarbon column. Net pay can only be interpolated after this has been done. Volumetrics are addressed by grid cell and then aggregated for each reservoir unit. For an oil reservoir unit, the volumetric algorithm can be written:

  Eq. 2

where STOIIP is the stock-tank oil initially in place and, for each grid cell, BRV is the bulk rock volume; N/G₂ is the net-to-gross reservoir fraction; φ₂ is the average porosity over the net-reservoir interval(s); Sh₂ is the computed hydrocarbon saturation over the net-reservoir interval(s); and B is the formation volume factor.

The summation of BRV across all grid cells equals the GRV. In computing the BRV, the base of the system should be taken to be a hydrocarbon-water contact, where present. Depending on project maturity, the recovery factor could be derived from simulation based on the geocellular model.

Although the use of net-reservoir vis-`a-vis net-pay cutoffs is partially self-compensating though different parametric averages, the introduction of geocellular models does lead to different estimates of STOIIP by the very nature of the process, and the potential impact of this cultural change needs to be quantified. In other words, in computing average properties over net-pay intervals, Eq. 1 excludes the water leg and the lowermost part of a transition zone as well as perched water intervals and zones of high capillarity. On the other hand, Eq. 2 can include all of these intervals in the parametric averaging process, subject to their satisfying net-reservoir criteria. With this in mind, the benefits of using synergic cutoffs, which reduce the disparity between net-to-gross reservoir and net-to-gross pay, are evident.

Contemporary methods of 3D reservoir modeling can accommodate greater reservoir complexity in the form of net-to-gross reservoir and porosity distributions, and also saturation vs. height variability. Several key stages can be identified in the context of integrated reservoir studies. Figure 3 maps an approach to using net pay for a deterministic geocellular application, although it can easily be adapted for geostatistical models. It assumes that net-pay criteria have initially been established using core data.

Fig. 3. Workflow illustrating the use of net reservoir and net pay in a geocellular volumetric approach to the estimation of petroleum resources in conventional reservoirs.

Other uses of net pay are to evaluate infill-drilling potential, to target zones for formation stimulation, to identify perforation intervals, to aid in the interpretation of well-test data, to guide the design of fluid-injection programs, to initialize reservoir simulators more effectively, to sharpen reserve estimates, and in equity redetermination, which is often based on in-place volumes and for which the procedures are usually proprietary.

RECENT TRENDS

Recent trends in net-pay analysis range from the generic meaningfulness of net-pay cutoffs to the role of net pay in horizontal-well applications, in thin-bed evaluation and in tertiary recovery programs. For example, a study published in 2010 compared different methods of estimating net-pay cutoffs in thick beds that can be resolved by logging tools.6 The research showed that regression methods, which assume an idealized dataset from a statistical perspective, are sensitive to departures from that ideal. Statistical methods that assume nothing about data character produce significantly different results. Clearly, these findings support the contention that the identification of cutoffs should always be driven by the data character.

Net-pay criteria have been used to select those thinner but log-resolvable reservoirs with bottom water that were suitable for horizontal well development.7 Interestingly, the net-pay concept was not transposed to the horizontal section of a wellbore, but instead was maintained in vertical space. This is the more meaningful procedure. Over the horizontal section, “net pay” can be increased simply by drilling further, and this improperly subordinates nature to oilfield operations.

One recent deepwater study compared net-pay analyses over thinly bedded reservoirs where the individual beds were below logging tool resolution.8 The comparisons showed that conventional formation evaluation furnished net-pay thicknesses that were considerably lower than those derived from thin-layer analysis. The results were not groundtruthed to core, which would be a prerequisite for exporting the methodology. Yet, the authors confirm how much pay has been missed in these thinly bedded sequences, which globally account for a good deal of bypassed pay.

Net pay has also been used as a screening parameter in thermal recovery programs applied to heavy-oil clastic reservoirs.9 Net-pay thickness must attain a certain threshold before a heavy-oil reservoir can be a candidate for thermal recovery. Higher recovery is typically associated with higher net-to-gross pay.

Another topic that has received attention recently is the role of net pay in predicting well performance in shale gas reservoirs. Although net-pay concepts are at an immature stage in their application to unconventional reservoir systems, there are preliminary indications that net pay can be used as a design parameter in fracturing programs.10 Much will depend on how net pay in unconventional reservoirs is defined and quantified. 

ACKNOWLEDGMENT
This article, with the permission of the Society of Petroleum Engineers, updates material previously published as Worthington, P. F., “Net pay: What is it? What does it do? How do we quantify it? How do we use it?” SPE Reservoir Evaluation & Engineering, 13, No. 5, 2010, pp. 812–822.

LITERATURE CITED
1 Worthington, P. F. and L. Cosentino, “The role of cutoffs in integrated reservoir studies,” SPE Reservoir Evaluation & Engineering, 8, No. 4, 2005, pp. 276–290.
2 Cosentino, L., Integrated Reservoir Studies, Editions Technip, Paris, 2001.
3 Worthington, P. F., “The application of cutoffs in integrated reservoir studies,” SPE Reservoir Evaluation & Engineering, 11, No. 6, 2008, pp. 968–975.
4 Ringrose, P. S., “Total porosity modelling: Dispelling the net-to-gross myth,” SPE Reservoir Evaluation & Engineering, 11, No. 5, 2008, pp. 866–873.
5 Amaefule, J. O., Altunbay, M., Tiab, D., Kersey, D. G. and D. K. Keelan, “Enhanced reservoir description: Using core and log data to identify hydraulic (flow) units and predict permeability in uncored intervals/wells,” SPE 26436 presented at the SPE Annual Technical Conference and Exhibition, Houston, Oct. 3–6, 1993.
6 Bouffin, N. and J. L. Jensen, “Efficient detection of productive intervals in oil and gas reservoirs,” Journal of Canadian Petroleum Technology, 49, No. 3, 2010, pp. 58–63.
7 Jin, H. et al., “Horizontal well production technologies in the thin-bedded, bottom-water reservoirs in Luliang Oilfield, China,” SPE 131890 presented at the International Oil and Gas Conference and Exhibition in China, Beijing, June 8–10, 2010.
8 Capone, G. et al., “Integrated formation evaluation in an anisotropic reservoir offshore deep water Indonesia using a combination of image logs, WFTs and mini DSTs for pay and hydrocarbon definition,” SPE 134076 presented at the SPE Annual Technical Conference and Exhibition, Florence, Italy, Sept. 19–22, 2010.
9 Lu, X. G., Sun, S. Q. and J. Xu, “Application of thermal recovery and waterflood to heavy and extra-heavy oil reservoirs: Analog knowledge from more than 120 clastic reservoirs,” SPE 130758 presented at the International Oil and Gas Conference and Exhibition in China, Beijing, June 8–10, 2010.
10 Kazakov, N. Y. and J. L. Miskimins, “Application of multivariate statistical analysis to slickwater fracturing parameters in unconventional reservoir systems,” SPE 140478 presented at the SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, Jan. 24–26, 2011.