Important Learnings for Reliable Management of Hydrocarbon Production and Salt Solution Mining induced Subsidence from Case Histories in the Netherlands
Abstract
Reliable management of subsidence caused by hydrocarbon production and salt solution mining is important for a country like the Netherlands where most land surface is below or near sea level. However, a factor two difference between prediction and observation is not uncommon. To nevertheless ensure a high probability that subsidence is kept within the limits an area can robustly sustain, a tightly integrated prediction/monitoring/updating loop is applied. Prior to production, scenario's spanning the range of parameter and model uncertainties are generated to calculate possible subsidence outcomes. The probability of each scenario is updated over time through confrontation with measurements (e.g. using Bayesian statistics) as they become available. Production can thus be halted or adjusted timely if probabilities start to indicate an unacceptable risk of exceeding set limits now or in the future. A number of projects with well documented, high quality prediction and monitoring were started in the Netherlands in the second half of the previous century. They provide quality case histories covering multi-decade production periods from which important learnings have been been extracted. Firstly, from the data it is clear that sandstone reservoir compaction is not a linear function of pressure depletion. Initially the rock in the field compacts much less than expected based on standard lab measurements. As pressure drops further, compaction gradually increases, reaching and exceeding lab values. Various mechanisms could be responsible: delayed compaction in lower permeability/poorly connected parts of the reservoir or aquifers; intrinsic non-linear, time-dependent, rate-type or diffusive behavior of the reservoir rock; previous deeper burial or increasing overpressure over geological time. The observed field behavior is described reasonably well by a single exponential time decay model. The non-linear and/or time-dependent field behavior has to be accounted for when updating predictions based on early field data. Otherwise it leads to under-prediction of subsidence, followed by multiple upward adjustments as new data become available. Secondly, the large difference between lab and field loading rate results in late time field compressibilities that can be 20 to 30% higher then the lab data. For chalk reservoirs the difference in loading rate causes much earlier pore collapse in the field. These effects need and can be accounted for. Thirdly, the case histories show that the shape of the subsidence bowl changes over time. The bowl shape becomes steeper in time for hydrocarbon extraction and flatter in the case of salt extraction. This is believed to be related to the changing elasticity contrast between the compacting volume and its surroundings as the reservoir compressibility increases and surrounding salt layers start to creep. The observed shape changes can be modeled numerically or by a varying rigid basement depth in the analytical van Opstal model. Not accounting for it can result in large subsidence allocation errors where salt mining and hydrocarbon production bowls overlap.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2012
- Bibcode:
- 2012AGUFMNH33B..07W
- Keywords:
-
- 5100 PHYSICAL PROPERTIES OF ROCKS;
- 4315 NATURAL HAZARDS / Monitoring;
- forecasting;
- prediction;
- 4323 NATURAL HAZARDS / Human impact;
- 4329 NATURAL HAZARDS / Sustainable development