Progressive Cavity Pumps (PCPs) are the predominant form of artificial lift method deployed by Australian operators in Coal Seam Gas (CSG) wells. With over five thousand CSG wells [
It is possible to gauge the holistic production performance of PCPs with the aid of real-time data, as this allows for pro-active and informed management of artificially lifted CSG wells. Based on data obtained from two (2) CSG operators, this paper will discuss in detail how features extracted from time series data can be converted to images, which can then aid in autonomously detecting abnormal PCP behavior.
Behrenbruch, Peter (Bear and Brook Consulting) | Hoang, Tuan G (University of Adelaide) | Do Huu, Minh Triet (Bear and Brook Consulting) | Bui, Khang D (Bear and Brook Consulting) | Kennaird, Tony (Bear and Brook Consulting)
Dynamic reservoir simulation models are used to predict reservoir performance, and to forecast production and ultimate recovery. Such simulation models are also used to match historic production. The success of such models depends critically on optimal gridding, particularly vertical definition and the choice of rock parameters, especially relative permeability.
This paper compares simulation results as a function of utilising alternative relative permeability relationships as simulation input:
Unaltered laboratory data Modified Brooks-Corey ( Relationships based on the more recently derived two-phase Modified Carman-Kozeny (
Unaltered laboratory data
Modified Brooks-Corey (
Relationships based on the more recently derived two-phase Modified Carman-Kozeny (
For maximum clarity, comparisons are made on a single layer basis but covering a range of permeability and porosity values, and capillary pressure relationships are based on modelled lab data using the Modified Carman-Kozeny Purcell (
Study results show that very different production responses may be realised, depending on the validity of original lab data and choice of modelled relationships deployed. It is concluded that the use of the
Capillary pressure (CP) measurements on core samples generally deploy one of three tests: porous plate (PP), centrifuge (CF) and mercury injection (MI). The first two mentioned tests are most often used to directly determine a CP relationship and the PP procedure most closely conforms to reservoir situations. CF test results may be superior for certain types of plugs and testing conditions, not to mention that testing is considerably faster. MICP tests are typically used for determining pore size distributions but may also be used to generate CP curves, with best results obtained by validation against the other two tests. MI tests also have the shortest timeline.
CP curves have been modelled (fitted or matched) over the years with many alternative formulations, from the well-known J-function approach (
This paper examines several CP data sets, comparing PP and MI results, in light of pore throat distributions. Seven formulation are used to match entire CP profiles from PP data and are subsequentlyexaminedfor being able to represent capillary entry pressure. Various types of pore structures are discussed: homogeneous (narrow distribution), broad (poorer sorting) and bimodal (two distinct distributions) where the smaller pores are often related to pore-fill. It is shown that the Modified Carman-Kozeny Purcell (MCKP) formulation (
Hydraulic Flow Zone Unit (HFZU) analysis for characterising cored formations using the Carman-Kozeny (C-K) equation was first proposed in the late 1980s. More recently this formulation was amended as for diverse formations, the C-K equation typically works in less than 50 percent of the cases. The modified formulation includes a cementation factor to handle C-K non-compliant cases. The objective of this paper is to build on these formulations, covering the full range of geological situations.
The study presented compares traditional reservoir characterisation approaches with the latest geological zonation techniques, showing how the full range of relationships may be universally accommodated. Such zonation is the best preparation for optimal reservoir simulation. In particular, it is shown how variable grain size and sorting may be visualised. Results are presented in traditional space, logarithmic permeability vs. porosity, and in model space: hydraulic radius, or reservoir quality index vs. porosity group, porosity fraction divided by the solid fraction.
It is shown that the methodology presented can deal with any sandstone (clastics) situation: C-Kcompliant and non-compliant, further categorising variable cementation and other diagenetic features, and including interbedded intervals and fining sequences. The zonation of a geological interval using the methods presented is the best basis for choosing vertical gridding for a dynamic reservoir simulation model, circumventing traditional upscaling, often fraught with inconsistencies. It has been found that the Flow Zone Indicator, a measure of grain size (or pore throat size) is the best parameter for zonation, followed by sorting of grains or pore throats. A Global Characteristics Envelope, encompassing 6 dimensions, is used to further validate final results.
The paper presents several case histories, covering fluvial and marine environments and a range of geological depositions and facie, including pore-fill (typically kaolinite). Fields are located offshore Australia and overseas.
Capillary pressure relationships (drainage) is of importance in both, reservoir engineering and petrophysics. Reservoir engineers use capillary pressure relationships in reservoir engineering calculations and petrophysicists use such relationships in saturation-height modelling. The overall objective of the study presented is to perform a comparative study of various capillary pressure models commonly used by the petroleum industry and to access their ability to match diverse laboratory data.
Over the years many capillary pressure formulations have been proposed to match laboratory data, from the well-known Leverett J-function approach to more recent methods. Generally, formulations may be theoretical, semi-empirical, empirical or statistical basis. The comparative study presented involves the following analytical formulations: modified Leverett J-function, modified Brooks-Corey (MBC), Thomeer, Skelt-Harrison, Lambda (λ) and the more recently established, Modified Carman-Kozeny-Purcell (MCKP) model.
Various types of pore structures may be modelled: homogeneous (narrow), broad (poorer sorting) and bimodal (two distinct distributions) where one of the distributions is often related to pore-fill, or a mixture of distribution types. Results presented in this paper compare application of the above-mentioned methods for homogeneous plugs. It is demonstrated that the MCKP method consistently outperforms the other methods and it is the only method that is able to identify the type of pore throat distribution without any further input data. The advantage of the MCKP method is that it does not contain any fitting parameters. The degree of accuracy of the MCKP model is typically R2 > 0.99. The MCKP model also lends itself very easily to predict capillary pressure relationships from more basic parameters.
Presented are examples with data from two Australian fields, one offshore and one onshore.
Capillary pressure relationships (drainage) involve key SCA parameters measured in the laboratory. Reservoir engineers use capillary pressure relationships in reservoir engineering calculations and dynamic reservoir simulation and petrophysicists use such relationships in saturation-height modelling. The objective of this paper is to describe a novel technique to identify pore throat or grain size distributions from capillary pressure profiles. Capillary pressure, laboratory data are traditionally obtained from three different types of experiments: centrifuge, porous plate and mercury injection. The modified Carman-Kozeny Purcell (MCKP) model may be used in matching various laboratory results. Initial application of the MCKP model involved a standard matching procedure without differentiating according to the type of pore throat or grain size distribution. In this paper a refined, more powerful analysis technique is described in detail where matching of any laboratory data with the MCKP model is obtained in model space, showing various characteristics. Characteristic profiles observed in model space can be directly attributed to the type of pore throat or grain size distribution: homogeneous (narrow), broad (poorer sorting) and bimodal (two distinct distributions), or a mixture of distribution types.
Determination of original hydrocarbon in place, OHIP, is a vital task in petroleum development. The estimation of appropriate rock and fluid properties is a requirement, with irreducible water saturation, Swir, being one of the key parameters. A representative Swir is also required when conducting multi-phase experiments, for example relative permeability determination. Furthermore, Swir has an influence on residual oil saturation during tertiary recovery.
Conducting primary drainage capillary pressure experiments to measure Swir is the industry-established practise. Such experiments tend to require considerable resources and long time periods to complete. As a consequence, a limited number of representative core plugs are typically considered, often leading to data gaps for some facies within a reservoir. In such situations, an empirical model may be useful in predicting Swir.
In a recent study, artificial neural networks have been applied successfully to the estimation of irreducible water saturation (Swir) for Australian formations. The model demonstrates a superior performance when compared with other, conventional models. This paper features the translation of the artificial neural network model into a simple mathematical equation that is suitable for quick hand calculation. Moreover, a new semi-empirical model to predict Swir is presented, containing five adjustable constants. The optimal values for these constants were obtained by minimizing the calculated error utilizing a genetic algorithm. Both neural network and semi-empirical models were developed, calibrated and validated by using an extensive data set gathered for Australian hydrocarbon basins.
The accurate determination of irreducible water saturation, (Swir), is important in reservoir engineering and petrophysical calculations, for example:
It should be mentioned that in petroleum engineering irreducible water saturation (Swir) and initial water saturation (Swi) are sometimes used interchangeably. This fact may be due to a misuse or misunderstanding of terms. The capillary pressure vs. saturation curve generated from a capillary pressure experiments, after translation to reservoir conditions, defines the saturation profile across a reservoir, and in an ideal situation corresponds to the actual saturation profile encountered in the reservoir (obtained from log analysis), representing connate (or geological) water. A laboratory profile, representing the initial water saturation profile may thus be transformed into a saturation versus height relationship. Swir, on the other hand, is a limiting value that is taken as a characteristic value for a particular reservoir. Considering a capillary pressure curve, Swir is the lowest water saturation value corresponding to the maximum capillary pressure value, and should be equivalent to that encountered at the reservoir crest.
The prediction of relative permeability has been in the past and is currently a very active research area, with theoretical, experimental and empirical approaches under consideration. However, it is fair to say that the complexities of relative permeability have to date eluded researchers and practitioners alike, in that there is no universal formulation that is able to predict two-phase relative permeability for the wide range of rock and wettability characteristics observed. This paper presents a new, generalised formulation, one that is truly predictive, and compares the same with the industry standard - the modified Brooks-Corey (MBC) formulation.
The MBC formulation is perhaps the most widely used, practical method describing laboratory-derived relative permeability relationships in terms of simple power functions. The shortcomings of this formulation are that it has no real predictive capability and the relative contributions due to pore structure components as compared to variation in wettability cannot be resolved. The new two-phase flow formulation presented is based on a phenomenological approach related to the Carman-Kozeny equation, and is able to resolve the above mentioned shortcomings.
Included are several laboratory examples and the results of a comparison of the two formulations is presented. It is shown how the new formulation is able to predict "the curvature?? of relative permeability curves when only the endpoints are known, duplicating observed behaviour from steady-state relative permeability experiments. Alternatively, if the endpoints can be derived, correlated or estimated with the use of more fundamental data, the entire prediction of relative permeability is possible.
In conclusion, the formulation presented is able to predict two-phase relative permeability under steady-state conditions, not just merely fit data. The second advantage of this method is that it is theoretically based and does not involve any fitting parameters but involves relatively simple analytical expressions.