38 |
Oilfield Technology
August
2015
criticality events, this level of evaluation is sufficient. The highest risks
should, however, be quantitatively assessed to avoid two common biases
to qualitative assessments:
Ì
Ì
Individual perspective.
Ì
Ì
Unrelated but critical events.
To reduce individual perspective bias, the risk assessment should
includemore than one subject matter expert whowill have different
perspectives. Next, the root cause of complex failures should be considered
as a series of multiple circumstances, whichmay or may not apply to the
current systembeing evaluated.
To highlight bias fromunrelated critical events and consequences, an
example used is that of theMacondo blowout, which occurred through a
series of failures related to cementing and subsea BOP function. If the risk
assessment goal is to evaluate single and dual barrier dry tree production
riser systems, then theMacondo event is not a good ‘likelihood’ reference.
The blowout and following oil spill would not have been prevented by a
dual barrier system. Instead, for the riser margin example given above, a
more appropriate likelihood should be used, such as failures per service
hour due to fatigue, connector leaks, riser disconnect, etc.
To reduce bias and increase confidence in assessment of these complex
scenarios, event trees as shown in Figure 3 are recommended. Event trees
allow:
Ì
Ì
Assessment of multiple outcomes from common failures or events.
Ì
Ì
Application of specific initiating event frequencies towards a defined
failure scenario.
Ì
Ì
Determination of gate probabilities.
Ì
Ì
Determination of consequence bands for specific outcomes.
With a basic event tree structure in place similar to that in Figure 3,
the process can be iterated for a number of riser failuremodes and time
durations inwhich the well is live (as shown in Table 1). The probabilities
are then summed down each tree branch, providing total event
probabilities for the risk assessment as given in Table 2. This approach
increases confidence and accuracy in the risk assessment, allowing for
better decisions between single and dual cased systems.
TheMacondo incident does, however, provide a good reference to
estimate the impact, or consequent cost, of a blowout occurring. Recent
estimates provide costs of approximately US$10 000/boe spilled. Other
large events have not been so costly.
1,6,7
The amount of barrels spilled and
anticipated cost/bbl can be assessed based on factors such as location, well
flow rates and gas versus liquid composition of the well.
CalculatingtheRiskex
With the risk events adequately defined alongwith amost likely outcome
for each, the next step is to calculate the risk costs for each system. The
‘value’ or Riskex of each risk event is the likelihoodmultiplied by the
consequence cost.
1,2
An example is provided in Table 3. In this example,
a blowout is assumed not to bridge over and spill 15 000 bpd until relief
wells can be drilled in approximately threemonths. The spill rate should be
developed for the specific site and a likelihood related to the event can be
derived from the event tree (Figure 3). There are two points to keep inmind
when determining the consequence.
Facilitydowntime
Facility downtime can be a driver for Riskex, but is only loosely related to
the number of riser casings. Consider the case where an integrated drill
rig is available on the platform to pull and repair a riser later in life (when
fatigue ismore likely). The downtime cost will be significantly less than
if a drilling rigmust be contracted and installed onboard before drilling
operations can commence. For instance, a single well producing 3000 bpd
at US$60/bbl costs approximately US$6million in deferred production
for a singlemonth’s downtime. The deferred production cost escalates
to US$36million considering sixmonths to contract a rig and complete
the repair. Therefore, a key decision to consider in consequence cost is the
ability to stop a flowingwell or complete a repair quickly.
General assumptions
The second point to bear inmind, is that general assumptions can
significantly affect the probability of occurrence as compared tomore
specific understanding, thus artificially altering the risk value. For instance,
a development with 15 wells can adopt one of the following assumptions:
Ì
Ì
A maximum of 10 sidetrack drilling operations for all wells.
Ì
Ì
A maximum of two sidetracks drilling operations per well.
Reference likelihoods are often given in terms of a per‑event probability
(e.g. probability of blowout of 1.0E‑6 per sidetrack completed). Therefore,
the total likelihood of a blowout for 10 planned sidetracks is 1.0E‑5, which
is three times less likely than the ‘two sidetracks per well’ yielding a
probability of 3.0E‑5. The total Riskex will also be three times higher due to
themore general assumption.
Finally, the Riskex is summed down for both the single and dual
casing riser systems as shown in Table 3. In the example, there is a delta of
approximately US$2million in Riskex between the two systems. The total
cost of each riser systemwould need to be estimated based on the same
assumptions used for the risk assessment. The cost should include both
riser hardware as well as the impact to vessel payload. Then the sumof
Capex, Opex plus Riskex for each systemcan be compared.
Conclusion
The approach described in this article is to help operators quantify the
risks and give adequate justification to a decision between dual and
single casing risers. In a post‑Macondo environment, engineers should
rely on a quantitative andmethodical approach to evaluate inherent risks
involved and demonstrate they are as lowas reasonably practicable. The
following key points are derived froma reviewof the risk assessment
process for single and dual cased risers:
Ì
Ì
Dual casing risers provide an extra barrier against the loss of riser
margin and other failures. The reduction in risk comes at an upfront
cost to the project, both of which can be quantitatively evaluated with
the methods outlined.
Ì
Ì
Failure probabilities and consequence costs depend on many factors.
Therefore, the entire operational philosophy should be considered
alongside relevant failure modes. The quality of a risk assessment
depends heavily on the quality of the inputs.
Ì
Ì
Given the limited number of riser failures to date, a quantitative
approach to some risks will be necessary. The cumulative likelihood
of a large blowout has not increased significantly, even though the
memory of such an event like Macondo is still fresh. However, recent
events have shifted the expectation of associated consequence costs.
Ì
Ì
Given the above considerations, a valuable risk assessment can only
be completed with the input of those familiar with riser systems and
their operations; drilling, completion and intervention operations; risk
assessment; and the specific field operating philosophy.
Ì
Ì
The benefits of a good risk assessment reach far beyond the results
of the assessment. A shift in operational philosophy derived from a
review of the site specific threats and potential mitigations can reduce
both the Riskex and Capex.
References
1.
Goldsmith, R., Eriksen, R. and Deegan, J., ‘Lifetime Risk‑Adjusted Cost Comparison for
Deepwater Well Riser Systems.’ OTC 10976 (1999).
2.
J.C. Dethlefs, SPE, ConocoPhillips; B. Chastain, Arktis, ‘Assessing Well Integrity Risk: A
Qualitative Model’, SPE‑142854, (April 2011).
3.
Goldsmith, R., ‘Risks and concerns of single‑casing riser systems versus dual‑casing
risers’; Offshoremagazine, (July 2011).
4.
Oil and Gas Producers (OGP) Risk Assessment Data Directory Report 434‑2,
Blowout Frequencies, (March 2010).
5.
Worldwide Offshore Accident Databank Statistical Report, 1997, DNV, Høvik, Oslo.
6.
Joint Industry Project, ‘Risk Assessment for Dry Tree Tieback Alternatives’, Phase 2
Study Final Report, (March 1998).
7.
Abel WildWell Control, L WAbel of WildWell Control Inc., ‘Blowout Risks Cut with
Contingency Plan’ Oil and Gas Journal, (7 June 1993).
