Gas Resource Estimation
Most sorbed gas reservoir systems are comprised of multiple lithologies having variable characteristics and resource concentrations. The methodology and terminology described here is intended to provide a single unified approach that encompasses both sorbed gas and free gas, is applicable to any lithology, avoids unwieldy ‘conversion factors’, and is readily adaptable to stochastic modeling.
What the world needs now, is….
…..more acronyms!! Here are a few, all of which mean the same thing:
It’s a gas!!
it?) In the definitions above, the term gas
actually refers to “natural gas”
, which is defined by SPE (2011) as:
“the portion of petroleum that exists either in the gaseous phase or is in solution in crude oil in natural underground reservoirs, and which is gaseous at atmospheric conditions of pressure and temperature. Natural Gas may include some amount of non-hydrocarbons.“
In other words, classification as “gas” is based on its physical state at surface conditions, not in the reservoir.
Although the definition doesn’t specify, the definition of natural gas can be inferred to include other non-gaseous condensed forms as well. SPE (2007) defines coalbed methane, for example, as “natural gas contained in coal deposits whether or not stored in gaseous phase. […]”. The same argument would be relevant to shale gas as well, which occurs partly in condensed form in the ground.
Calculation of GIIP has become unnecessarily complicated, particularly through the usage of a somewhat awkward set of measurement units. The calculation can be substantially simplified when it’s slightly rearranged, as follows.
GIIP represents the sum of two components:
[Total] GIIPt = GIIPs (Sorbed Gas Initially In Place) + GIIPf (Free Gas Initially In Place)
GIIPs and GIIPf are not formal terms, but serve a useful purpose, as they are calculated separately. For water saturated coals, GIIPf would be nil, and for organic matter-free tight sands, GIIPs will be effectively nil, but most shale gas reservoirs will include contributions of both. The terms simply drop out of the calculation if one of the parameters is zero.
Gas Resources are the fraction of GIIP that is actually recoverable, calculated by multiplying GIIP by a Recovery Factor (RF). Owing to the curvilinear shape of (most) sorption isotherm(s), however, RF will be different for sorbed gas and free gas. Thus, different recovery factors should be used if both phases are initially present.
Producible Sorbed Gas Resource = GIIPs x RFs
Producible Free Gas Resource = GIIPf x RFf
Many resource estimations simply avoid this question for shale gas by assuming that RFs
equals zero; however, this may be significantly incorrect, depending on bottom hole flowing pressure, fluid pressure gradients within the formation, and abandonment pressure. Estimation of recovery factors is unavoidably one of the most difficult and uncertain aspects of resource estimation, and requires an understanding of past performance of analogous reservoirs, anticipation of the potential impact of emerging drilling and production technologies, and a valid model of reservoir mechanisms.
Some resource estimations use different resource equations for different reservoir types: tight sandstone, carbonaceous shale, and coal. It makes more sense, however, to consolidate both GIIPs and GIIPf into a single unified resource equation that is applicable to any lithology by choosing an appropriate set of parameter values. This approach: 1) offers greater consistency–and, therefore, less chance for error, 2) has a smaller number of parameters overall, and 3) is very amenable to Monte Carlo simulation.
GIIPs and GIIPf can be readily combined by calculating the “Equivalent Volume Ratio” (EVR) for each, where EVR represents the number of volume units of GIIP (normalized to STP), present in an equivalent volume of reservoir. As the same units are used in numerator and denominator, EVR is effectively dimensionless. Something like acre-feet could be used as a measure of reservoir volume, but the corresponding GIIP will be in acre-feet as well… not a very practical unit.
. Free Gas Equivalent Volume Ratio is calculated as:
EVRf = Φgi x Egi
where Φgi is the volume fraction of the reservoir initially occupied by the gas phase and Egi is the initial “Expansion Factor”, which is the ratio of the gas volume at STP to its volume in the reservoir, calculated from the real gas law (pV=ZnRT). Both terms are dimensionless.
Resource estimation procedures frequently use a different parameter, Formation Volume Factor (Bg), rather than Egi, to correct gas volumes to STP conditions. Bg is usually defined as the ratio of the volume (of gas) in the reservoir to volume (of gas) at the STP; however, in some cases the inverse is intended. What’s more, in some cases (Bg incorporates a 1000X multiplier (e.g. Zuber, 1996). Rather than have to deal with these potential sources of ambiguity and error, Expansion Factor is used here instead, with no multiplier. The meaning of “Expansion Factor” (as a ratio) is more self-evident than “Formation Volume Factor” which, although it has long been in use in reservoir engineering, is not .
The parameter Φgi provides a direct estimation of the volume fraction of the reservoir occupied by gas phase. This would include the combined sum of matrix porosity and fracture porosity. The only reason to treat them separately in the resource calculation would be if their recovery factors were different.
The corresponding term in other resource estimation procedures is the product of two parameters: Total Porosity (ΦT, and Initial Water Saturation (Swi). This is an approach that has been applied in conventional reservoirs, but does’t translate well down to microporous rocks. The significance and measurement of porosity at nanometer scale is uncertain, and water saturation is difficult to measure accurately adding another parameter into the calculation multiplies the uncertainty in Monte Carlo modeling. Arguably, gas-filled porosity can be guessed with as much total porosity and water saturation can be measured. Estaimation comes down to a combination of labortory data, geophysical log data, and reservoir behavior to come up with a reasonable estiamte.
EVRs. For the sorbed gas fraction:
EVRs (Sorbed Gas Equivalent Volume Ratio) = SGC x RhoB
where SGC is the sorbed gas content, on a whole rock basis, and RhoB is the bulk density of the reservoir rock.
To calculate GIIP:
GIIPf = EVRf x Reservoir Volume
GIIPs = EVRs x Reservoir Volume
GIIPT = [EVRf + EVRs] x Reservoir Volume
To yield a useful value, the reservoir volume (area x thickness) should be measured using the same volume units as are desired for the resource estimate, e.g. cubic meters or cubic feet.
Accounting for Reservoir Variability. Sorbed gas reservoir systems are compositionally heterogeneous—both stratigraphically and laterally, particularly where multiple lithologies contribute toward the total resource. Sorbed Gas Content and gas Expansion Factor both vary as a function of temperature and pressure, which vary with depth. Sorbed gas content is also strongly controlled by organic matter content, which in turn has a strong impact on bulk density. Sorting out the interrelationships among these parameters is one of the challenges in attempting to maximize the accuracy and precision of resource estimates, and one where I believe I can bring significant added value to resource estimate.
There are two possible approaches to account for variability in reservoir quality: deterministic or statistical (stochastic).
Deterministic Methods. In the deterministic approach, a single value is estimated for each of the parameters in the resource calculation. If desired, different scenarios can be defined representing a high (optimistic) case, middle case, and low (conservative) case.
A simple implementation of the deterministic method can be used to estimate GIIP for CBM based on representative estimates of coal seam thickness and total gas content. (See Utility Lookup table here.)
The deterministic approach can also be implemented in a more rigorous and detailed manner, however, by subdividing the reservoir stratigraphically and/or regionally into component elements, each having its own set of parameter values. The resource estimates for each element can be summed to yield a total.
Deterministic approach to regional assessment of GIIP for a dipping stratigraphic sequence comprising multiple coal seam reservoirs of varying depth and gas content
Example #1: Regional assessment of CBM resources in dipping strata in northern Colombia comprising multiple thick coal seams. Gas content for each seam varied substantially, related to depth. The entire proprety was divided into polygons representing the surface outcrop of a particular stratigraphic interval containing one or more seams. Each polygon included stratigraphic units progressively higher in the section in the down-dip direction. GIIP for each polygon was estimated by summing the contribution of each stratigraphic subdivision.
Example #2: I have developed a methodology for a geophysical log-based deterministic estimate of total gas resources for complex heterogeneous stratigraphic sequences that include coal, shale, and sandstone. The number of reservoir elements is limited only by the resolution of the log data. The resulting calculations may be used to generate an useful representation of the stratigraphic distribution of gas resources, that can be used in economic modeling or in designing drilling & completion programs.
Custom Log Analysis Procedure, picking coals and carbonaceous shales for integrated estimation of GIIP over the full stratigraphic section
Example #3: Another very powerful implementation of the deterministic resource calculation that I have developed utilizes correlations among variables to generate regional grids for the parameters used in the resource calculation, each grid being representative of a particular stratigraphic unit occurring over a large area. Many grids are required, as there may be multiple steps involved in generating the required parameters. Then, utilizing “grid math” functionality of a regional mapping program, these grids can be combined to create a grid representing the three dimensional distribution of GIIP throughout a project area. This method is particularly useful in that it can be applied to strata occurring over a wide range of depths, having variable thermal maturity, thickness, organic matter content, etc.
Statistical (Stochastic Modeling) Approach. Stochastic modeling uses random combinations of parameter values to generate a statistical range of possible outcomes. Stochastic modeling of gas resources entails three steps:
- Rather than using a single value for each parameter, define a statistical range of input values. This entails selecting an appropriate statistical distribution function that best approximates the natural variability of the parameter being estimated, as well as a range of parameter values based upon the function selected. (See note below.)
- Define statistical interrelationships, or dependencies, among parameters, so that impossible combinations of values will not be generated in the subsequent step
- Conduct Monte Carlo modeling, to randomly generate a complete set of parameter values through thousands of iterations, which yields a statistical range of possible outcomes. The results can be used to determine low (P90), median or “best” (P50), and high (P10) case estimates, representing the 10th, 50th, and 90th percentiles, respectively, which are used in the PRMS resource classification scheme.
(IMPORTANT NOTE ON STATISTICAL NOMENCLATURE: In order to conform to PRMS protocol, statistical distributions for both the input parameters and results of Monte Carlo simulations must be specified according to their 10th, 50th, and 90th percentiles in the statistical distribution. Unfortunately, in conflict with international statistical nomenclature, SPE refers to the 10th percentile as “P90″ and the 90th percentile as “P10″, which could potentially cause confusion. SPE also uses the term “Best” in reference to the 50th percentile, or median case (which thankfully is unambiguously labeled “P50″!). Thus, Best doesn’t actually mean “best”, as in the expression “best case scenario”, but something more along the lines of “most realistic”, as in the expression, “What’s your best guess?”. Specifying certain types of statistical distributions, particularly lognormal, can be problematic based on P90, P50, and P10.)
A Role for Geological Understanding. Commonly, a “triangular distribution” is used represent the input range parameters, usually because it is the simplest distribution to define. Here, a better understanding of the nature of the system can lead to a more accurate model. For example, bed thickness, and sorbed gas content, which may be better modeled by a lognormal distribution.
Defining dependencies among parameters is an important step in Monte Carlo simulations, as without them, the predicted outcomes will include unlikely, if not impossible, results. Defining these dependencies, however, requires a strong understanding of reservoir mechanisms, and the complex interrelationships among variables.
Resource Classification. Based upon criteria specified in the PRMS classification system, resource estimates can be classified into three major categories: Reserves, Contingent Resources or Prospective Resources based upon the likelihood of commercial development, then further subdivided into additional classes based on technical (largely geological) uncertainty.