ESTIMATING DRILLING COSTS-1: Joint association survey, mechanical risk index methods common in GOM

Aug. 6, 2007
Over the past several decades, various methods have been proposed to evaluate drilling costs and complexity, but because of the large number of factors and events that affect drilling performance, predictive models have been difficult to construct.

Over the past several decades, various methods have been proposed to evaluate drilling costs and complexity, but because of the large number of factors and events that affect drilling performance, predictive models have been difficult to construct.

The joint association survey (JAS) and mechanical risk index (MRI) are the most popular methods used to evaluate drilling costs and complexity in the US Gulf of Mexico, and specialized indices have been introduced to characterize the complexity of drilling directional and extended reach wells.

Quantifying well costs and complexity is complicated by restrictions on data collection and availability, constraints associated with modeling, or combinations of these factors. Drill rates are often constrained by factors that the driller does not control and in ways that cannot be documented.

Recently, the concept of mechanical specific energy has been used to improve bit efficiency and performance and obtain a more objective assessment of drilling efficiency.

The purpose of this three-part series is to review the primary methods used to assess drilling costs and complexity and to propose a new method that combines the best features of each approach.

Part 1 describes the JAS and MRI methodologies. Part 2 introduces metrics that characterize directional and extended reach wells, along with the concept of mechanical specific energy, which has seen recent success. Part 3 introduces a generalized approach to drilling cost estimation.

Drilling factors

Drilling a hole in the ground in search for or production of oil and gas is subject to significant sources of variability. Although the physics of drilling is the same everywhere throughout the world, geologic conditions, contractor experience, equipment availability, well specification, and various other factors can lead to a wide range in drilling performance. Cost estimation is difficult and benchmarking efforts are often unreliable.

Performance comparisons are mostly done on a well-by-well, actual-vs.-plan basis or seek to correlate costs to performance indicators, metrics, or drilling parameters. Evaluating the differences that exist in drilling wells and comparing costs requires establishment of statistically reliable relationships between performance metrics and factors that impact drilling.

Formation geology at the site and location of the target reservoir are primary factors that influence drilling cost. Geologic formations vary across the world and within the same producing basin. Hard, abrasive, and heterogeneous formations typically have low penetration rates, frequent drillstring failures, and significant deviation from the planned trajectory.

Deep reservoirs are usually characterized by low permeability, high temperature and pressure, complex fracture growth and stress regimes, and contaminants such as CO2 and H2S that increase the complexity of the well and require operators to deal with a number of issues concerning safety and operational performance.

Drilling methods used to make hole depend upon geologic formation and technology applied, amount of information known about the formation, experience and preferences of the operator, available equipment, and the drilling contractor’s experience and execution.

Characteristics of the well are specified by drilling plan, location of the target reservoir, and conditions encountered during drilling. Bit hydraulics has a major influence on drilling efficiency; its role is complex because it is closely tied to other drilling variables, such as lithology, bit type, downhole conditions, mechanical drilling parameters, circulation system and drilling mud.

Site characteristics such as water depth, operator’s experience in the region, and expected environmental conditions influence the operator’s selection of contract and rig type, which in turn influence drilling performance metrics. Exogenous events such as stuck pipe, adverse weather, and mechanical failure cannot be predicted and can have a significant impact on the time and cost to drill a well.


Two methods are commonly used to benchmark drilling performance.

The first method is based on experimental design and controlled field studies. Typically, one or more parameters of the drilling process are varied and the impact of the variable(s) on output measures such as the rate of penetration (ROP) or cost/ft examined.1

One of the most common approaches is the “drill rate” test, in which the driller experiments with various weight on bit (WOB) and rotations/min (RPM) settings, and selects the parameters that result in the highest ROP. Controlled field studies are often the best way to understand the relationships between drilling factors under a set of conditions that are tightly controlled.

The analytic results derived under field studies are often based on engineering and scientific principles specific to the wellbore conditions, experimental design, equipment, and contractor, and so the ability to generalize and apply the results to other wells and locations may be limited.

The second method to study factor effects is based on an aggregate assessment of well data collected from various contractors, locations, and well bores.

In this method, data that characterize a set of wells are collected and relationships are established between the variables based on empirical modeling techniques.2 The aggregate approach to analysis uses a set of drilling data and seeks to discover relationships between various factors of drilling and the cost and complexity of the wellbore. It is common to try to capture the best practices by comparison to an ideal well or offsets.

For instance, the approach used in “technical limit” drilling describes a level of performance defined as the “best possible” for a given set of design parameters.3 This allows engineers to compare a variety of factors that impact drilling and to develop models that describe the behavior of the performance metrics.

Cost estimation

Since the drilling budget can represent a significant part of the capital expenditures for field development, drilling operations are carefully planned and closely watched, and operators maintain meticulous and detailed records of each well drilled.4

Cost estimation is performed specific to the drilling prognosis. The usual procedure is to decompose costs into eight general categories:

  1. Site preparation.
  2. Mobilization and rigging up.
  3. Drilling.
  4. Tripping operations.
  5. Formation evaluation and surveys.
  6. Casing placement.
  7. Well completion.
  8. Drilling problems.

Spreadsheet models are employed using various levels of detail. Typically, several categories are specified, with the drilling engineer itemizing the expected time and cost/category.5-7

Each cost component is identified and subdivided into minor cost elements, and the percentage contribution of the total cost for each major category is computed to help identify the key cost drivers. To improve the range of the estimate, the uncertainty of the cost drivers is frequently quantified.8 9 A contingency is added to accommodate some of the uncertainty of costs before the final authorization for expenditure (AFE) is determined. The well budget is then sent to management for approval.

JAS history

The joint association survey on drilling costs has been performed in the US since 1954 in cooperation with the American Petroleum Institute, Independent Petroleum Association of America, and Mid-Continent Oil & Gas Association. The first cost surveys were performed in 1944, but 1954 is generally recognized as the official start of the JAS. Since 1959, JAS data have been collected and published annually.

The purpose of the JAS is to provide information pertaining to drilling costs and expenditures for finding, developing, and producing oil and gas in the US. The JAS is the only publication in the US that contains annual state-by-state and offshore drilling cost data.

Questionnaires are mailed to operators to verify information on well completions performed during the year and to provide cost data for each well drilled. The response rate of the survey varies, but typically, 40-50% of operators respond to the request for information, representing 40-60% of the total number of wells and footage drilled during the year.

Because not all operators respond to the survey, it is necessary to estimate drilling costs for unreported wells. The JAS accomplishes this task by constructing models to infer the expected costs of drilling for unreported wells. The model-estimated costs are added to the reported costs to obtain the total estimated expenditures for the year.

Primary variables

The geographic-offshore or onshore-location of each well is specified and the well type (exploratory, development) and well class (oil, gas, dry) designated. An oil well is defined as a well completed for the production of crude oil from at least one oil zone or reservoir, while a gas well is one that can produce hydrocarbons existing initially in gaseous phase.

The total depth of the well is the total feet of penetration drilled down the wellbore, including water depth and all plugged-back footage, but excluding bypassed footage from sidetrack drilling.

Well direction is classified as vertical or horizontal. Most offshore exploration wells are drilled vertically, while typically only the first development well is vertical; subsequent wells are drilled vertical to a certain depth and then kicked off to target. The majority of onshore footage is vertical, while total offshore footage is primarily directional.

Wells are evaluated after the drill bit reaches the target depth. A drillstem test may be used to evaluate the flow rates of hydrocarbons, and integrating these data with logs and other tests leads to the completion decision.

Click here to enlarge image

The total cost of a dry well includes the cost to set concrete plugs and to remove casing, as required by local, state, and federal regulations. The total cost of a producing well includes the cost through completion and installation of the Christmas tree.

Completion costs will typically include the cost of casing and production tubing, perforation, packers, safety devices, kits at the reservoir sands (e.g., gravel pack, wire-packed screens), and a tree at the top of the well.


The JAS cost-estimation procedure evolved in five distinct phases:

  • 1954-65: Wells were classified according to geological structure, drilling conditions, and economic expectations. Well cost/ft drilled by depth range was regressed against the average depth/well in each class interval for each region and well class for both tangible and intangible costs.10
  • 1966-77: The average cost/ft drilled was computed for wells classified according to well type, location, and depth.11 The tangible and intangible cost categories were aggregated and regression lines computed to describe the functional relationship between cost/ft and depth for each area under consideration as shown in Equation 1.
  • 1978-92: A stepwise linear regression on the cost/ft for each sample area and well type was employed.12 Three depth variables were applied: inverted depth, depth, and depth squared, as well as a set of dummy classification variables for well type (oil, gas, dry), well class (exploratory, development), and completion type (single, multiple).

    Equation 2 specifies the function; coefficients αi (i = 0, ..., 3) and βi (i = 1, ..., 9), are estimated through least-squares regression. I_i ( i = 1, ... 9) are indicator variables for each of the nine wellbore classification categories.
  • 1993-94: Regression models were developed for well type and geographic area with the functional relation of Equation 3, where α, β0, β1, and β2 are determined by regression.13 A “stabilizing” transformation was performed by adjusting α to convert the dependent variable to a form that was linearly correlated with the independent variables. Three transformations were found to be statistically significant: the natural log, α = 4, and α = 0.5. The estimates were then adjusted with a correction factor to eliminate the bias introduced by the transformation.
  • 1995-present: Wellbore data are currently aggregated into 16 geographic regions following the Gas Research Institute’s hydrocarbon supply model.14-15 A nonlinear two-factor regression model is constructed for each region based upon the model specification described by Equation 4.

The X1 and X2 variables are numeric, while the X3, X4, and X5 variables are categorical, defined in terms of indicator variables; e.g., (0, exploratory well; 1, development well).

The coefficients αi (i = 0,1, ..., 5) and αij (i, j = 1, ..., 5, i


In the JAS drilling cost model, four variables-total depth, well type, well class, and well direction-are applied in a two-factor, nonlinear regression model. Two-factor interaction terms were incorporated in the model to “build up” the number of available terms and improve the statistical fit of the regression.

The limitations of the procedure are obvious from the model construction because four variables cannot possibly describe the complexity and operational aspects involved in drilling a well except on an average aggregate gross basis. A quadratic expression is appropriate for the requirements of the survey, but it is clear that the JAS methodology cannot provide a reliable cost predictor on an individual well.

Pipe stands ready in the derrick of Noble Drilling Corp.’s Amos Runner semisubmersible, drilling in Green Canyon Block 955 in December 2006 (photo by Nina M. Rach).
Click here to enlarge image

The JAS procedure is successful for estimating (unreported) cost data and reporting expenditure patterns, but to predict individual well cost the level of categorization is too broadly defined to be useful except as an average.

A single well is characterized by a large number of descriptor variables which are not captured in the survey response, and thus, inadequately represented in the output model. A more robust model would incorporate additional descriptor variables of the wellbore and drilling process and relax the quadratic specification.

MRI history

The mechanical risk index was developed in the late 1980s by Conoco engineers charged with comparing offset drilling data for a collection of offshore wells in the Gulf of Mexico.16 They developed a “mechanical risk index” to compare operations and derived an algorithm based on empirical analysis of well data, taking into consideration factors such as water depth, measured depth, and kickoff point for sidetracks.

In the mid-1990s, Dodson modified the MRI using key drilling factors, copyrighted the formula, and incorporated the measure as part of a commercial well database ( Reference to the MRI is found in various trade publications,17 but little systematic analysis of the metric has been performed.

The MRI is defined in terms of four “component factors” and a weighted composite “key drilling factor.” The component factors are described in terms of 6 primary variables and the key drilling factor represents the composite impact of 14 qualitative indicators. The MRI is computed as an additive function of the component factors weighted by the composite key drilling term.

Primary variables

The six primary variables of the MRI include the total measured depth (TD), vertical depth (VD), horizontal displacement (HD), water depth (WD), number of casing strings (NS), and mud weight (MW) at total depth.

The depth of a well measured from the rotary table along the length of the wellbore is called the total depth (or total measured depth), while the (true) vertical depth is the distance from the rotary table measured in a vertical plane to TD. The horizontal displacement is the distance measured in plan view from the rotary table to TD. The water depth is the distance from the waterline to mud line.

The problems, costs, and hazards of drilling depend upon both observable and unobservable factors. The deeper the hole, the more time is lost in round trips to replace worn bits and to run casing, tests, logs, etc. As the depth of the well increases, the number of formations encountered typically increases along with the number of casing strings required to maintain well control.

As the number of casing strings increases, the trip time, installation, and cementing time will increase, all increasing drilling time and cost. Beyond a certain depth, technical complications and the opportunity for problems increase drastically.

Casing serves several important functions in drilling and completion and is one of the most expensive parts of a drilling program, making up 10-20% of the average cost of a completed well.6 A well that encounters no abnormal formation pore pressure gradients, lost-circulation zones, or salt sections usually requires only conductor and surface casing to drill to the target.

Deeper wells that penetrate abnormally pressured formations, lost-circulation zones, unstable (sloughing) shale sections, shallow water flows, or salt sections generally will require one or more strings of intermediate casing to protect formations and to prevent well problems.

The number of casing strings in a well provides an indirect measure of well complexity because complex wells are frequently associated with multiple strings, and narrow margins between formation pore pressure and fracture pressure gradients often require a greater number of casing strings.18

Wells are usually drilled with water or oil-based muds through the entire wellbore or one mud may be displaced for another over a selected interval. The mud weight at total depth serves as a proxy for the wellbore formation pressure. With all other factors equal, the greater the hole pressure, the heavier the mud, and the slower the drilling.

Underbalanced drilling requires the use of special equipment to handle formation fluids entering the well; its primary use has been where casing is set and cemented on top of a subnormal or pressure-depleted formation (OGJ, Dec. 1, 2003, p. 39).19

Component factors

The primary variables of the MRI are combined into four normalized component factors shown by Equations 5-8. Each component factor is nonlinear in the primary variables.

Drilling factors

Key drilling factors are defined to capture drilling characteristics that are encountered, or are expected to be encountered, but not described by the component factors. Dodson introduced drilling factors to generalize the MRI to a larger class of wells.

The key drilling factors are user-defined qualitative variables that are assigned an integer-valued weight according to the occurrence of the condition and its degree of complexity. The composite key drilling factor is the sum of the drilling factor weights given in Equation 9, where Ψi denotes the ith drilling factor of well w and Ψi(w) is the assigned weight.

Variables and corresponding weights are defined in the Nomenclature box.

Most exploratory wells are drilled as straight as possible, while usually only the first development well is vertical. Horizontal drilling is less stable than vertical drilling and the wells are more difficult to log and complete.20 If a horizontal section is drilled, then a weight of “3” is assigned to the key drilling factor, while if a J-shaped or S-shaped trajectory is employed, an additional weight of “3” or “2” is included in the metric.

A subsea well is one in which the wellhead, Christmas tree, and production-control equipment are located on the seabed. Subsea well drilling tends to be more complicated and costly than a normal tree installation, and a weight of “2” is assigned to subsea completions.

Hydrogen sulfide and CO2 environments require special operating procedures. Hydrogen sulfide is a poisonous and corrosive gas that causes embrittlement and weakening of steel casing and drill pipe. Wells with high CO2 concentrations also suffer corrosion problems.

One of the technical problems in deepwater drilling is the formation of hydrates in the blowout preventer (BOP) or choke and kill lines. Hydrates can plug the BOP stack and well-circulation path and are difficult and time consuming to remove.

Mature fields have intrinsic drilling problems associated with their depleted reservoirs. The water-wet sands that typify depleted zones propagate seepage losses and differential sticking. Drilling-fluid losses are frequently unavoidable in large fractures, and pressured shales are often interbedded with depleted sands, requiring that multiple pressure sequences be stabilized with a single drilling fluid.21

Ductile salt is effective for trapping oil and gas because it can move, surround, and deform sediments, creating traps. Drilling salt is risky, however, because salt is weak and undergoes continuous deformation.22 Sediments below salt are often disrupted and overpressured, and constructing long-lasting wells through salt requires special fluid selection, casing programs, and cementing procedures.

A slimhole well describes a borehole significantly smaller than standard and commonly less than 6-in. or 6½-in. in diameter. Mud-hole suspension systems and coring also add to the time and complexity of drilling.

Unusual geologic and environmental events, such as loop currents, eddies, and shallow water hazards, create special problems during drilling. Loop currents and eddies subject facilities to stress and vibration, and drilling risers that are in place may bend or bow from the current to such an extent that the vessel has to change position to stay connected.

In some cases, the drill pipe may rub against the drilling riser forcing immediate shutdown. Shallow water flow occurs during drilling into over-pressured sand zones.23 Installation of additional casing is usually required to maintain wellbore integrity in shallow flow.

MRI definition

The MRI is defined through the component factors, weighted by a normalized composite key drilling factor, shown in Equation 10.

The MRI is frequently used to compare the drilling performance of two or more wells and as a predictive tool for wells in the design stage. MRI is also correlated to drilling cost.

Click here to enlarge image

For a well that has previously been drilled, the input data and MRI can be calculated precisely. If a well is part of a planned drilling program, then estimates for the variables (TD, WD, VD, HD, MW, NS) and key drilling factors (Ψ1, ..., Ψ14) are required to estimate the anticipated drilling risk. An example MRI calculation is shown in the box on this page.


The MRI was originally developed to compare the drilling performance of a small number of offset wells drilled in the late 1980s. As such, the formulation of the metric is closely associated to the characteristics of a particular well set drilled during a specific period of time. The MRI was later modified to incorporate additional drilling factors not covered in the original formulation.

The MRI currently serves as the de facto industry standard in the Gulf of Mexico. It has a long history, is easy to comprehend and useful in aggregate comparisons, and is defined by simple, spreadsheet-programmable relationships. The parameters of the MRI are based on a minimal set of high-quality drilling data that are readily acquired. Thus, there is much to recommend the MRI.

Several issues associated with the metric, however, deserve closer attention. Dodson introduced drilling factors to generalize the MRI to a larger class of wells, but selection of factors and their weight assignment appear arbitrary. The use of the drilling factors serves to create a cost-estimation tool, but the manner in which the parameters enter the model as a binary indicator with weighting factors may lead to ambiguity.

The application of user-defined weights is always problematic. If weights are not inferred through an empirical assessment of well data, the assignment can be considered arbitrary and may possibly be ambiguous; e.g., if a horizontal section of a well is drilled, a weight of “3” is assigned to the key drilling factor. On a cost/ft basis, however, horizontal wells are not necessarily more expensive than vertical holes.15

Key drilling factors are assigned weights according to the “complexity” of the characteristic or condition that is encountered (or expected to occur) without differentiating between the magnitude of the condition; e.g., if a horizontal section of a well is drilled, then regardless of its length, it is assumed to be three times more complex or difficult than if a salt section is drilled or if a loop current is encountered. A loop current that leads to a 3-hr delay is treated the same as a 3-day delay.

Primary and key drilling factors represent the drilling process in a manner superior to the JAS variable selection, but the manner in which the MRI factors are combined and the weight selection can be improved. The composite drilling factor weight is ad hoc, and it would be better to normalize the component factors prior to summation.

Although the MRI incorporates more drilling parameters than the JAS approach, the JAS methodology is more structured, and it is clear that the manner in which factors are incorporated in the MRI limits generalization. The MRI is defined by an additive functional and a fixed weight selection. Generally speaking, metrics defined through a formula assignment are not expected to be optimally specified.


  1. Bourgoyne, A.T., Millheim, K.K., Chenevert, M.E., and Young, F.S., Applied Drilling Engineering, Society of Petroleum Engineers: Richardson, Tex., 2003.
  2. Iyoho, A.W., Meize, R.A., Millheim, K.K., and Crumrine, M.J., “Lessons from integrated analysis of GOM drilling performance,” SPE Drilling and Completion, Vol. 20, No. 1 (2005), pp. 6-16.
  3. Bond, D.F., Scott, P.W., Page, P.E., and Windham, T.M., “Step change improvements and high rate learning are delivered by targeting technical limits on subsea wells,” SPE/IADC 35077, 1996 SPE/International Association Drilling Contractors Drilling Conference, New Orleans, Mar. 12-15, 1996.
  4. McCammon, K.C., and MacKinlay, W.M., “The development and implementation of a drilling database: a case study,” SPE 26257, SPE Petroleum Computer Conference, New Orleans, July 11-14, 1993.
  5. Saifaldeen, Abdul Jabbar M., and Thabet, Sayed Abdul Rahman, “Integrated footage contract vs. day rate: drilling time and cost analysis,” SPE/IADC 39255, SPE/IADC Middle East Drilling Conference, Bahrain, Nov. 23-25, 1997.
  6. Jenkins, P.B., and Crockford, A.L., “Drilling costs,” SPE 5266, SPE-European Meeting, London, Apr. 14-15, 1975.
  7. Noerager, Jere A., White, J.P., Floetia, A., and Dawson, R., “Drilling time predictions for statistical analysis,” SPE/IADC 16164, 1987 SPE/IADC Drilling Conference, New Orleans, Mar. 8-11, 1987.
  8. Peterson, Susan K., Murtha, James A., and Schneider, F.F., “Risk analysis and Monte Carlo simulation applied to the generation of drilling AFE estimates,” SPE 26339, SPE Annual Technical Conference and Exhibition, Houston, Oct. 3-6, 1993.
  9. Kitchel, G.B., Moore, Steven O., Banks, W.H., and Borland, Brett M., “Probabilistic drilling cost estimating,” SPE 35990, SPE Petroleum Computer Conference, Dallas, June 2-5, 1996.
  10. 1964 Joint Association Survey of the US Oil and Gas Producing Industry, Section I: Drilling Costs, American Petroleum Institute: Washington, June 1967.
  11. 1974 Joint Association Survey of the US Oil and Gas Producing Industry, Section I: Drilling Costs, API, March 1976.
  1. 1985 Joint Association Survey on Drilling Costs, API, November 1986.
  2. 1993 Joint Association Survey on Drilling Costs, API, November 1994.
  3. 1995 Joint Association Survey on Drilling Costs, API, November 1996.
  4. 2001 Joint Association Survey on Drilling Costs, API, December 2002.
  5. Williams, Chad, Mason, John S., and Spaar, John, “Operational efficiency on eight-well sidetrack program saves $7.3 million vs. historical offsets in MP 299/144 GOM,” SPE/IADC 67826, SPE/IADC Drilling Conference, Amsterdam, Feb. 27-Mar. 1, 2001.
  6. Dodson, James, and Dodson, Ted, “Drilling efficiency numbers static,” Offshore, Vol. 55 No. 9 (January 2003), pp. 26-28.
  7. Proehl, Thomas S., “Pore pressures, fracture gradients, and drilling economics,” IADC/SPE 27493, IADC/SPE Drilling Conference, Dallas, Feb. 15-18, 1994.
  8. Bourgoyne, Adam T., “Well control considerations for underbalanced drilling,” SPE 38584, SPE ATCE, San Antonio, Oct. 5-8, 1997.
  9. Joshi, Sada D., “Cost/benefits of horizontal wells,” SPE 83621, SPE Western Regional/AAPG Pacific Section Meeting, Long Beach, CA, May 19-24, 2003.
  10. Hariharan, P.R., and Judge, Bob, “Benefits of dual-gradient methods for drilling upper hole sections,” World Oil, January 2003, pp. 44-47.
  11. Barker, J.W., Feland, K.W., and Tsao, Y.H., “Drilling long salt sections along the US gulf coast,” SPE 24605, SPE ATCE, Washington, DC, Oct. 4-7, 1992.
  12. Bourgoyne, Darryl A., Bourgoyne, Adam T., and Hannegan, Don, “A subsea rotating control head for riserless drilling applications,” IADC/SPE 33523, IADC International Deepwater Well Control Conference, Houston, Aug. 26-27, 1998.

The author

Click here to enlarge image

Mark J. Kaiser ([email protected]) is research professor and director, research and development at the Center for Energy Studies, Louisiana State University, Baton Rouge. His primary research interests are policy issues, modeling, and econometric studies in the energy industry. Before joining LSU in 2001, he held appointments at Auburn University, the American University of Armenia, and Wichita State University. He holds a PhD in industrial engineering and operations research from Purdue University.