A study commissioned by the UK Health and Safety Executive in 2005 showed that well control incidents are by far the most frequently occurring of all high-pressure/ high-temperature (HP/HT) well safety concerns. According to the r

Figure 1. According to a 2005 UK Health and Safety Executive study, well control incidents are by far the most frequently occurring of all HP/HT well safety concerns. (Images courtesy of Knowledge Systems)
eport, “the main issue is the high-pressure regime that exists in the prospective zone and more specifically the narrow margin between the required borehole pressure to control the reservoir pore fluid pressures and the allowable pressure to retain borehole competence” (Figure 1).

Data limitations in UHT/HP wells

With pressures and temperatures as high as 30,000 psi and 500ºF (260°C), ultra HP/HT (UHP/HT) wellbore environments are beyond the limits of current logging-while-drilling (LWD) technology and at the upper limit of wireline tool specifications. Operators should anticipate that data from sonic logs, pressure-while-drilling and resistivity tools will either be unattainable, unreliable (high failure rates) or prohibitively expensive.

These wells must be planned, drilled and produced using significantly less formation data than their shallower and cooler counterparts. As a result, pre-drill estimates of key formation properties — pore pressure, in-situ stress and rock strength — must be relied on.

Pore pressure estimation without LWD

In the absence of LWD data, pore pressure estimates must rely solely on the accuracy of the pre-drill geopressure model, updated and recalibrated with information available from drilling processes alone. This alternative to LWD and wireline-based pore pressure estimation is illustrated schematically in Figure 2.

Figure 2. Pre-drill geopressure model, updated and recalibrated with information available from drilling process alone, is an alternative LWD and wireline-based pore pressure estimation.
Although fairly new, recent advances in geomechanical modeling are producing very promising results for real-time pore pressure and wellbore stability estimation without LWD data. Such operations are not unprecedented — typically, this has occurred where operators have chosen to eliminate the high cost of LWD service, or where LWD data was found to be unreliable.

The prediction process

There are three components to the pore pressure prediction process — basin study, pre-drill predictive model and real-time updating. In the basin study, a 3-D geomechanical earth model is developed that incorporates information throughout the basin (e.g., water depth, age, thickness, lithology and geothermal gradient). Calibrated and integrated with seismic to adjust for local anomalies such as faulting and salt diapers, the model is used to prepare a pre-drill pore pressure prediction for the entire well profile.

To verify this methodology, a Joint Industry Project (DEA119) was conducted in 2003-2004 for five basins in the Gulf of Mexico and the North Sea. For 62% of the wells, the predicted pore pressure was within 1 pound per gallon (ppg) over the entire length of the well bore. For 38% of the wells the prediction was within 0.5 ppg.

Real-time model updates

Effective support requires real-time updates of the model while the well is being drilled in order to predict pore pressure and stability issues ahead of the bit. Until recently, real-time pressure updates were not generally feasible because of the high level of computational power and subject-matter specialization required. However, recent advances that reduce model complexity allow analysis on a PC laptop, which has opened up modeling to personnel with a broader general knowledge of the realities of day-to-day decision-making in a drilling environment.

Personnel must understand the importance of the process and their roles. There must be clear communications between rig-based data providers, the pore pressure analyst, drilling engineering staff and rig-based decision-makers and implementers. It is necessary for pressure analysts to have a strong working knowledge of the rig environment — both for the equipment used and to communicate information in an unambiguous and timely manner.

Flawless performance of the data transfer and management systems is also critical. Collaborative industry standards such as WITSML (Wellsite Information Transfer Standard Markup Language) are important to ensure data accessibility and integrity.

Updating models without logs

The key to real-time modeling downhole pressure without LWD or wireline data is to start with a robust geomechanical model, then extract as much pressure-related information as possible from all available drilling parameters.

The useful data falls into three major categories: 1) drillstring mechanics, which provide information on hole conditions and rock strengths; 2) fluid system hydraulics, important because the mud system transmits information about in-situ pressures as it interacts with in-situ pore fluids; and 3) fluid system contents (cuttings, gas content, etc.), which provide a wealth of information about downhole lithology and pressure conditions. While more information is better, this methodology has been applied using only commonly available drilling data from drilling equipment meters and sensors, daily drilling reports and mud-logging systems.

Taken in isolation, much of this data is inconclusive. For example, a change in rate of penetration can indicate that a formation top has been encountered or possibly that bit performance has deteriorated because of wear. However, when the rate of penetration change is accompanied by other indicators such as an increase in sand content and a change in chloride count, there is sufficient information to support a model update for a formation top. Updating the model allows observed wellbore behavior to be used to improve the prediction in the next section to be drilled. This allows the operator to be proactive in decision-making.

With a few exceptions, all of the measures are indirect indicators of borehole conditions. Taken together, they form a robust set of information from which to build viable pressure estimations. In many applications, real-time updates to the model are significant enough to change the mud weight recommendation two to four times during the course of drilling.

Performance without LWD data

In a typical instance, a major oil company was drilling a 14,000 ft (4,268 m) well in the Gulf of Mexico with minimal offset well information. Concerned about possible well control and integrity problems, the drilling team commissioned a real-time pore pressure model that relied on LWD data to generate updates.

About 1,800 ft (548 m) above total depth (TD), the sonic and resistivity tools malfunctioned. The team was faced with a tough choice: a costly trip to replace the tools, or drill ahead with a higher risk of well control and integrity problems. It opted to drill ahead while continuing to update the look-ahead pore pressure model using well response and drilling data.

This new updating methodology resulted in significant changes in the mud program, which helped to drill the well successfully to TD without control or other pressure-related drilling problems. Subsequently, when the well was logged with wireline, the log-derived pore pressures were very close to the values predicted using drilling data to calibrate the 3-D geomechanical model.

Adequacy vs. accuracy

As with any analysis, the addition of other independent sources of information will improve results provided the information is not erroneous. The consideration here is whether the improvement in model accuracy warrants the additional cost of LWD systems.

One way to look at this issue is to consider the decrease in pore pressure estimation error as additional estimation capabilities are added. No estimation (e.g., using hydrostatic estimates only) could lead to mud weights being 5 ppg or more too high or low at points within the well.

With a pre-drill model, the error is typically reduced to a range of ±2-3 ppg, at a cost of US $50,000 to $100,000. Continuous model updating using drilling parameters can reduce uncertainty to ±1 to 1.5 ppg, at a cost of $20,000 to $50,000. With LWD data the error can be reduced to as low as 0.5 ppg, but at a significant additional cost, often over $1 million. If the safe operating window is +/- 1.5 ppg or greater, then the running of LWD tools to further refine the estimate is simply not warranted. The objective should be to select a methodology based on the adequacy of the estimate, not the accuracy.