Original scientific paper Received: August 3, 2015 Accepted: September 2, 2015 Stochastically improved methodology for probability of success (POS) calculation in hydrocarbon systems Stohastično dopolnjena metodologija računanja verjetnosti uspeha (POS) v ogljikovodičnih sistemih Tomislav Malvić1, 2 *, Josipa Velić1 1INA Plc., Field Development Sector, Šubićeva 29, 10000 Zagreb, Croatia 2University of Zagreb, Faculty of Mining, Geology and Petroleum Engineering, Pierottijeva 6, 10000 Zagreb, Croatia Corresponding author. E-mail: tomislav.malvic@zg.t-com.hr Abstract Geological risk of new hydrocarbon reserves discovering is usually calculated on deterministical or expert-opinion way, and expressed as 'Probability Of Success' (abbr. POS). In both approaches are included selections of single probability values for each geological event organised into geological categories that define hydrocarbon system. Here is described a hybrid, i.e. stochastical, model based on the deterministical approach. Here was given example from the Croatian part of the Pannonian Basin System (abbr. CPBS), improved with stochastically estimated subcategory for porosity mapped in the Stari Gradac-Barcs Nygat Field (Drava Depression). Furthermore, there is theoretically explained how such approach could be applied for two other subcategories - quality of cap rocks and hydrocarbon shows. Presented methodology could be advantageous in clastic hydrocarbon system evaluation. Key words: geological risk, determinism, stochastics, Neogene, Northern Croatia Izvleček Geološko tveganje odkritja novih zalog ogljikovodikov navadno računajo deterministično ali po metodi ekspertnih mnenj in ga izražajo z »verjetnostjo uspeha« (POS, Probability of Success ). Obe metodi temeljita na vrednostih verjetnosti posamičnih geoloških dogodkov, urejenih po geoloških kategorijah, ki opredeljujejo ogljikovodični sistem. Tu je opisan hibridni, tj. stoha-stični model, ki temelji na determinističnem načinu. Obravnavani primer je iz hrvaškega dela sistema Panonske kadunje (CPBS, Croatian part of the Pannonian Basin System), ki je dopolnjen s stohastično ocenjeno podkategorijo poroznosti, kartirano v polju Stari Gra-dac-Barcs Nygat (v Dravski kadunji). Sledi teoretska razlaga možnosti uporabe takega načina z nadaljnjima dvema podkategorijama - kakovostjo krovnih kamnin in ogljikovodičnih pojavov. Prikazana metodologija utegne biti učinkovita pri ocenjevanju klastičnih ogljikovo-dičnih sistemov. Ključne besede: geološko tveganje, determinizem, sto-chastičnost, neogen, severna Hrvaška Introduction Calculation of geological risk is a well-established tool for estimation of possible hydrocarbon reservoir in plays or prospects. Such calculations, in Croatia, are well described in the Sava and Drava Depressions [1-4]. The term 'play' in those papers is generally defined as a stratigraphical unit in the range of chronostrati-graphic stage or sub-stage where hydrocarbon production already exists. The 'prospect' is a vertical surface projection of potential reservoir lateral borders. Such definition has been derived from Rose [5] or White [6] where 'play' is generally defined as an operational unit characterised by several prospects and/or fields and 'prospect' is an exploration (economic) unit. In general, any potential hydrocarbon system can be evaluated with Probability of success (abbr. POS) calculation. Mathematically, calculation of POS is a simple multiplication of several, in most cases five, independent geological category probabilities. Of course, there are geological relations among some of them, but it is using this tool impossible mathematically expressed on any useful way. Each category is defined with several geological events, and each also has its own probability. Category probability is simple multiplication of selected event probability values, defined as discrete values in range 0-1. They are often listed in POS probability tables, based upon previous experience and expert knowledge from analysed subsurface. Such a table (Figure 1), defined through decades of research for the Croatian part of the Pannonian Basin System (abbr. CPBS), had been a source of detail probability values defined and applied in in the Bjelovar Subdepression as part of the Drava Depression. Sometimes such values remain as an TRAP RESERVOIR SOURCE ROCKS MIGRATION 1 HC PRESERVATION | Structural Reservoir type Source facies HC shows Reservoir pressure Anticline and buried hill linked to basement 1.00 Sandstone clean and laterally extended; Basement granite, geiss, gabbro; Dolomites and Algae reefs (secondary porosity) 1.00 Kerogen type I and/or II 1.00 Production of hydrocarbons 1.00 Higher than hydrostatic 1.00 Faulted anticline 0.75 Sabdstones, rich in silt and clay; Basement with secondary porosity, limited extending; Algae reefs, filled with skeletal debris, mud and marine cements 0.75 Kerogen type III 0.75 Hydrocarbons in traces; New gas detected >10 % 0.75 Approximately hydrostatic 0.75 Structural nose closed by fault 0.50 Sandstone including significant portion of silt/clay particles, limited extending; 0.50 Favourable palaeo-facies organic matter sedimentation 0.50 Oil determined in cores (luminescent analysis, core tests) 0.50 Lower than hydrostatic 0.50 Any "positive" faulted structure, margins are not firmly defined 0.25 Basement rocks, including low secondary porosity and limited extending 0.25 Regionally known source rock facies, but not proven at observed locality 0.25 Oil determined in traces (lumin, anal., core tests) 0.25 0.25 Undefined structural framework 0.05 Undefined reservoir type 0.05 Undefined source rock type 0.05 Hydrocarbon are not observed 0.05 0.05 Stratigraphic or combined Porosity features Maturity Position of trap Formation water Algae reef form 1.00 Primary porosity >15 %; Secondary porosity >5 % 1.00 Sediments are in catagenesis phase ("oil" or "wet" gas- 1.00 Trap is located in proven migration distance 1.00 Still aquifer of field-waters 1.00 Sandstones, pinched out 0.75 Primary porosity 5-15 %; Secondary porosity 1-5 % 0.75 Sediments are in metagenesis phase 0.75 Trap is located between two source rocks depocentres 0.75 Active aquifer of field-waters 0.75 Sediments changed by diagenesis 0.50 Primary porosity <10 %; Permeability <1x10**(-3) micrometer**2 0.50 Sediments are in early catagenesis phase 0.50 Short migation pathway (<=10 km) 0.50 Infiltrated aquifer from adjacent formations 0.50 Abrupt changes of petrophysical properties (caly, different facies) 0.25 Secondary porosity <1 % 0.25 Sediments are in late diagenesis phase 0.25 Long migration pathway (>10 km) 0.25 Infiltrated aquifer from surface 0.25 Undefined stratigraphic framework 0.05 Undefined porosity values 0.05 Undefined maturity level 0.05 Undefined source rocks 0.05 0.05 Quality of cap rock Data sources Timing Regional proven cap rock (seals, isolator) 1.00 1.00 Geochemical analysis on cores and fluids 1.00 Trap is older than matured source rocks 1.00 1.00 Rocks without reservoir properties 0.75 0.75 Analogy with close located geochemical analyses 0.75 Trap is younger than matured source rocks 0.75 0.75 Rocks permeable for gas (gas leakage) 0.50 0.50 Thermal modeling and calculation (e.g. Lopatin, Waples etc.) 0.50 Relation between trap and source rocks is unknown 0.50 0.50 Permeable rocks with locally higher silt/clay content 0.25 0.25 Thermal modeling at just a few locations 0.25 0.25 0.25 Undefined cap rock 0.05 0.05 Undefined data sources 0.05 0.05 0.05 Figure 1: Example of relevant database prepared for the Bjelovar Subdepression 'after2 internal document, but only if published [eg 2 3] they make possible further independent evaluation of local or regional petroleum systems. Oppositely, such general tables, which can be applied as a rule of thumb, are missed in case of expert opinion applied for each particular well, exploration or development plan (Figure 2). In such case, single expert or team are completely responsible for given category values. Consequently, such process is subdued to "heavy" benchmarking, i.e. corrections are done with each new dataset (especially from wells). This methodology is not discussed here. However, it is obvious that, using deterministic approach, at least several geological events (Figure 2) can be estimated from the range, i.e. from interval defined with values, number of data and, sometimes, measurement error. Moreover, in the case of low number of inputs, the Monte Carlo sampling can be applied for generation of artificial data, but it needs to be clearly stated in statistical results. However, the key question is "can any probability value for each geological event be considered certain or not". If there is a measurable uncertainty (Figure 2), resulting in non-representative mean or variance, but the minimum and max- imum could be approximated, the stochastics can be successfully applied, e.g., using 2nd introduction of uncertainty in cell-value estimations with sequential Gaussian simulations. Such application of stochastics and results are shown, for the CPBS, in estimation of the porosity, thickness and depth of hydrocarbon reservoirs [eg 7-9]. Similar approach obviously can be regularly applied for estimation of several events in POS calculation and eventually set up as standard part of that method. Selection of stochastically mapped porosity in POS calculation The hydrocarbon plays or prospects could be deterministically analysed by several, mathematically independent, geological categories. The most common are: (1) structures, (2) reservoirs, (3) migration, (4) source rocks and (5) preservation of hydrocarbons [eg 2 3 6 10]. The values of events in the most category values can be evaluated from data collected from well files, logs, seismic, cores, descriptive geological interpretations or the comprehensive regional papers [eg 11-13]. Based upon those data, Analysed here DETERMINISTICAL SYSTEM Purely deterministical system based onto large number of discrete values set for analysed geological system. Values are considered as constants, categories are independent. Advantages The procedure is clear and can be transparently report to anyone. Disadvantages Ask for large knowledge about analysed subsurface volume, often collected after decades of exploration. Not analysed here HUMAN SUPERVISED BENCHMARKING SYSTEM The humans use some discrete values based on own or community experience. The POS could be calculated from independent or depend events. Advantages Can be applied for ever subsurface volume, even poorly explored. Analogy can be applied. Disadvantages Completely based on expert knowledge, hardly can be independently re-checked. Methodology can be mathematically improved, respecting geological possibilities «Interval of allowed estimation uncertainty» can easily be applied in: (a) sequential mapping of some (sub)categories, (b) establishing interval of values from available measurements Figure 2: Deterministical vs. human dominant benchmarking in evaluation of hydrocarbon systems. a value from the probability table can be easily selected, if such exists for the explored area (like Figure 1) or even from analogy based on regional geological models, especially deposi-tional and tectonic data [eg.' 11]. In any case, POS table makes possible to calculate such value for any play or prospect in the area where it is defined by Equation 1: POS = p (structures) x p (reservoir) x p (migration) X p (source rocks) x p (preservation) (1) Where are: POS - probability value of Probability of Success for analysed hydrocarbon system, p - probability value of each considered geological category. All geological events, subcategories, categories and POS are defined with numerical values. For the part of them inputs (laboratory measurements, loggings tools ...) strictly define the results (like kerogen type, quantity of hydrocarbons during drilling) and probability can be selected without uncertainties. However, some subcategories like 'Porosity features' (in the category 'Reservoir'), 'Quality of cap rocks' ('Trap'], and 'HC shows' ('Migration'] can be calculated from cores, logs and diagrams, but very often as approximations. It means they includes uncertainties, but if lithology is well-known the minimum and maximum values (e.g., for porosity) could be clearly established. The methodology had been tested with porosity maps taken from the Badenian gas-condensate reservoir in the Stari Gradac-Barcs Nyu-gat Field [14]. The reservoir is of massive type, trapped with combined structural-stratigraph-ic closure, with very complex lithology divided in four lithofacies (but single hydrodynamic unit). Porosity is geostatistically mapped in the youngest lithofacies of the Badenian clastites. The porosity distribution corresponds with structural strike NW-SE [15], and maps had been calculated using 100 realizations of sequential Gaussian simulations. It means that each cell on the map is defined with minimum and maximum values (realization), as well as 98 others between them. All of them, as equally probable, had been summed and averaged. So the minimum (3.1 %), median (3.2 %) and maximum (3.53 %) average reservoir porosities are calculated, what was base for consequently calculation of three solutions for 'Original Gas In Place' (abbr. OGIP) volume [16]. Trap Structural 5tJ ei Cici га phi с or combined Reservoir Source rocks Reservoir type Source facies Porosity features Maturity Migration Preservation of HC Ut shows Reservoir pressure Position of trap Formation water Quality of cap rock Data sources Timing Tr)[b Gmalo^cmt r-iif-cgcirrfls Quality ■of ^p CccH- tfrsf ccnjId Лг indirètti? mtuii-iitaii Лас1яЛЛсяПу Stipulai Gcoi&pitàt subcategories C-PfJ'JTmtC