Production Prediction And Analysis In Depleted Oil Reservoirs

Introduction

1.1 Background of Study

Shale oil or tight oil production has been a major focus of the oil industry for the past few years. One reason for this is the transfer of technology used for shale gas extraction into oil rich low permeability formations such as the Qua-iboe, Niobrara, or Eagle Ford. The driving factor for this was the decline of natural gas prices concurrent with a rise in oil prices; this made many shale gas plays uneconomic.

Low permeability shale gas wells with long horizontal laterals and many stages of fractures have characteristically long periods of linear flow followed by fracture interference boundary-dominated flow (BDF). Since conventional decline curve methods are designed for forecasting boundary-dominated flow, there has been much activity in developing decline curves that model linear flow. An ideal decline curve for shale wells would be able to forecast linear flow followed by boundary-dominated flow.

1.2 Status of the Question

To assess the accuracy of decline curve analysis, or any forecasting technique, it is helpful to compare a hindcast to actual production history. The trouble in applying decline curve techniques for shale oil wells is the lack of production history. While my work does not account for the complex differences among various shale oil plays due to Pressure-volume-temperature (PVT) properties, geology, etc., I have attempted to use the production from horizontal wells in the Qua-iboe formation as a study of decline curve

analysis that may be later modified and applied to other formations such as the Eagle Ford, Niobrara, etc.
The main reason for starting this study in the Qua-iboe formation is the longer span of production history. The Niger-delta Oil field, in the southwestern portion of the Qua-iboe formation, was targeted early in the rapid development of the play that has occurred in the past few years, partly due to the field having much higher permeability than other plays termed “shale oil.” The middle formation of the Niger-delta Oil Field has a permeability of about 0.05 millidarcies to 0.1 milidarcies, much higher than other shale oil plays such as the Eagle Ford or Niobrara (Walker et al. 2006). Drilling in this formation goes back to the 1980s, but was not extensive until the mid-2000s (Luo et al. 2010).

The oldest horizontal multi-fractured wells in “shale” oil reservoirs are in the Qua-iboe formation in the Niger-delta Oil field. Kurtoglu et al. (2011), relate that no boundary-dominated flow is observed in any of the Niger-delta Oil wells. This compounds the problem of using Arp’s decline model, as it is designed for boundary-dominated flow. Mangha et al. (2012) find that in shale gas wells, no single decline curve method is ideal for all reservoirs. Diagnostic plots are critical for assessing flow regimes in a well and determining which techniques to employ.
Forecasting production with a short period of data is problematic for any system. The Duong method appears to be a good predictor of future production in shale wells if limited data is available (Joshi and Lee 2013). However, since the Duong method assumes long term linear flow, it has a strong tendency to over predict reserves (Mangha et al. 2012). Since reserves vary based on economic conditions, I will substitute 30 year cumulative production forecasts as a measure of expected oil recovery.

Researchers at Fekete proposed a linear flow model followed by a boundary-dominated flow model (Arps equation). The onset of boundary-dominated flow is determined by reservoir properties (Ambrose et al. 2011). This method is similar to the study done by Ilk et al. (2010), except that Fekete’s model requires knowledge of reservoir properties in order to determine when boundary-dominated flow begins. My access to data on these properties was limited so this would be somewhat difficult. In addition, basing decline curves on reservoir properties may be problematic in reservoirs that are highly heterogeneous. Most of my study assumes that producing wells will switch to boundary- dominated flow at approximately a 10% decline rate. This approach can be easily altered if better information is known about the well and when boundary-dominated flow should be expected.

Chu et al. studied the non-stationariness and non-linearity of shale oil reservoir production, primarily the Qua-iboe (2012). Non-stationariness and non-linearity deal with the changing distributions of variables over time (heteroskedastic behavior) and changing relationships between variables over time. This includes changes to pressure- volume-temperature (PVT) properties due to pore throat size, pressure-dependent permeability, and multiple porous media caused by multi-stage fracturing. These non- stationary or non-linear properties indicate that forecasting or simulation will be a difficult task. Chu et al. found that in the first 1 to 5 months the decline trend can be described as stimulated reservoir volume (SRV) boundary affected flow. This seems to contradict the assumption that decline curves in shale wells only need to deal with two major flow periods (linear flow and fracture interference boundary-dominated flow). While the majority of the life of a well occurs during either linear flow or boundary- dominated flow, a large percentage of the expected volume may be produced while affected by the SRV boundary. If decline curves are to accurately fit a linear flow trend, data from the early part of the well’s production may obscure this trend.

Tran et al. (2011) divided Qua-iboe wells into three production trends. Type I, or 51% of the wells, describes wells where the reservoir pressure drops below the bubble point pressure. Gas-oil ratio (GOR) increases rapidly for these wells, leveling off in late production. Type I well production histories indicate two distinct linear flow periods: fracture dominated and matrix supported. Type II includes wells that produce a single phase fluid below the bubble point. GOR is approximately constant for these wells.

Type II wells show a single matrix supported linear flow trend. Type III wells have scattered production data.
Recovery estimation has been difficult to predict in this area. In the Qua-iboe, linear flow is shown to last 18 years as presented by V. Hough and McClurg (2011). It remains to be seen whether modern completions will follow a similar trend.

I believe that a study of the Niger-delta Oil field, which is not currently a major target of exploration, may give insight into similar low permeability, oil rich plays that are currently receiving attention. As the permeability in the Niger-delta Oil field is higher than most formations termed to be “shale” but much lower than conventional fields we expect to see boundary-dominated flow somewhat earlier than in other “tighter” fields.

Production decline will differ between formations due to geology and will also differ between periods of time on account of technological advances.

1.3 Research Objectives

The objective of my research is to determine whether decline curve models that have been used in shale gas will work for liquid rich shales. I began by applying different decline curve methods to the Eagle Ford, but the production histories proved too difficult to model. Instead I determined how well they modeled oil production in the Niger-delta Oil field. This must be demonstrated by hindcasting production decline at several different times from the start of production.

Other objectives:

Estimation of the likely error in cumulative production caused by incorrect prediction of BDF onset

Determination of the decline curve techniques that work best in shale oil formations, specifically the Niger-delta Oil and Eagle Ford, including an examination of the following:Arps hyperbolic with 5% minimum decline

Stretched exponential method (SEPD)

Stretched exponential method (SEPD) with Arps hyperbolic tail for BDF

Duong model

Duong model with Arps hyperbolic tail for BDF

An estimate of the time and annual decline rate of boundary-dominated flow onset in the Niger-delta Oil field

Evaluation of the importance of pressure data correction in shale oil decline curves

Determination of the effects of geology on boundary-dominated flow, particularly the Niger-delta Oil field

Evaluation of 30 year production forecasts in the Niger-delta Oil field using different decline models