Not Just Drilling Optimization

SPE-181027-MS
Using RealTime Data for Well Design Optimization, Not Just Drilling
Downloaded from http://onepetro.org/SPEIE/proceedings-pdf/16IE/All-16IE/SPE-181027-MS/1411918/spe-181027-ms.pdf/1 by United Energy Pakistan Ltd user on 11 September 2023
Optimization
Øystein Andersen, AGR Software
Copyright 2016, Society of Petroleum Engineers
This paper was prepared for presentation at the SPE Intelligent Energy International Conference and Exhibition held in Aberdeen, United Kingdom, 6-8 September 2016.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
In recent years, amount of downhole drilling dynamics data has increased significantly. To some extent, the
velocity of which the data becomes available for evaluation / correction has also increased, especially with
the introduction of wired drill pipe. But even without this technology, it is evident that the amount of data
one can evaluate to optimize performance, has increased substantially from only a decade ago.
There is a large focus on adjusting drilling parameters to enhance drilling progress / ensuring wellbore
stability, but experience shows that the data tends to be "forgotten" after a well is finished. This is partly
due to the volume of data, but mostly the engineering team's inherent trend to design wells again from
scratch based on scarce "lessons learned" spreadsheets from the previous well and theoretical simulations;
simulations which often have large uncertainty built-in.
Contradicting many people's belief, drilling engineering is not an exact science. The operations take place
several thousands of feet below the ground and one only get hints of what is happening downhole. It is up
to the engineering department to make as much out of these hints as possible, often referred to as "listening
to the hole" in order to ensure a successful drilling operation.
With lots of parameters affecting the drilling performance (such as e.g. mud properties, bit/BHA design
and trajectory), there is no clear indication that one can find the optimum strategy for a drilling a hole
section by changing only one factor between one well and the next. Even so, changing drilling parameters
to optimize hole performance during operation is only a part of the optimization, it is in the planning phase
one makes the decisions that can really affect if a well will be successful or not.
For a field development (or any other area with several wells being nearby each other) it is recommended
to change the method of designing wells. Instead of leaning towards sophisticated theoretical simulations,
one should instead have a much higher focus on having the historic data "at your fingertips" when designing
a new well. But in order to do this analysis, it is crucial that the data available is of high quality. This means
having log data in similar format and well details & operational timings entered correctly in the daily drilling
reporting system.
It is, however, not enough only to compare the drilling parameters / well designs, one also needs to assess
what operations were successful/unsuccessful for each wellbore together with the relevant data. Examples
of this can be annotation of formation instability with mud parameters or annotation of bit damage together
with drilling parameters.
2 SPE-181027-MS
From offset well information, one can utilize a probabilistic approach for well design. This includes what
risk factors to consider and expected ranges operational timings. From this, one can easily see where the
largest risk factors and opportunities for improvements are.
Introduction
In today's market in the oil & gas industry, the focus on cost has become extremely high. This also affects
the drilling and well operations, where companies are to a large degree optimizing drilling performance.
Drilling parameters are modified constantly in order to increase the on bottom ROP, but in most cases we
see that this overall does not have a very high significant impact on the well's total performance, typically
on bottom drilling only accounts for ~40 % of the time from spud to TD.
Overall well construction performance is affected by many other factors, many of which are needed to
address long before the BHA goes through the rotary table. Examples are BHA design, mud types and casing
point selections. Making these decisions correctly may lead to an enhanced on bottom drilling performance,
but to a much larger extent they will minimize flat times on time vs depth curves. Flat times in this context,
means time being spent without making further progress on the well.
Make the most of historic data

There are two types of flat times we are referring to; planned flat times (such as casing points) and unplanned
flat times caused by NPT (Non-Productive Time). Planned flat times should be reduced as much as possible
(we should therefore highlight what operations the time is spent on) and unplanned flat times should be
eliminated as much as possible.
In order to reduce / eliminate flat times it is of crucial importance to estimate them as well as possible
in the planning stage of a well. Many say that this is an impossible task as one never knows what is going
to happen, and therefore choose to add a percentage of NPT on top of the time estimate. In our opinion,
however, this is possible when utilizing a probabilistic approach.
With a probabilistic approach it is possible to assign ranges (both in time and cost) to planned operations,
expected drilling speeds etc., but also expected consequences if an incident (i.e. an unplanned flat spot)
occurs. The range for the consequence should be determined from historic data. If you can investigate offset
occurrences of being stuck with a casing and not being able to pull it and see that the NPT delay ranges
from 2 to 5 days, then this is probably a good range to set as the consequence if this situation should happen
on your planned well. But in order to have such an approach, it is off course necessary to have this historic
database of downhole occurrences available, and it must be intuitive to ensure that you are comparing
"apples to apples".
The consequence of the downhole event happening is one side, but in order to assign risk one also need
to estimate the likelihood of such an incident happening. Again, having the historic data available at the
fingertips is what allows finding a sensible likelihood, example: 1 in 5 wellbores with similar designs in
similar area have gone stuck with 9 5/8" casing → 20% likelihood of getting stuck.
As for risk factor's likelihood and consequence it is also possible to have other historical data available at
the fingertips in order review various tool's performance and from this evaluating whether it should be run
again or not. This is to some extent done in various ways in companies, but in our opinion not to the extent
that one has the ability to. Historic well experiences are often kept within the teams instead of residing with
the company as a whole. For example, it is in our opinion not beneficial to lean on a hydraulic simulation
in order to estimate pump pressure at the end of a section, it's much better to actually review what the pump
pressure was at TD for a historical comparable well.
The ultimate goal with setting up a probabilistic time and cost model is to highlight in which phases the
largest contributors to delays are, both in expected timings but also where potential risk lies hidden so that
the well can be severely delayed by NPT. These slow (or expensive) operations can there be optimized as
SPE-181027-MS 3
much as possible and risks may be mitigated as much as possible. Residual risk will always be present to
some degree, but at least it can be highlighted for management.
As we see, the following captured data from a historic well needs to be QCed, standardized and made
available "at the fingertips" together with general well data (well location, rig utilized, date drilled etc.) and
intuitively visualized so that it can be made beneficial for comparison when planning a new well (possibly
several years later):
– Experiences (downhole drilling events, learnings with specific tools)
– Drilling Parameters (ROP, WOB, flow etc.) coupled together with relevant experiences such as
vibration damage, unable to steer BHA etc.
– Geological data (pore pressures, LOT/FITs, composite logs)
– Mud parameters (MW, ECD, rheology) coupled with downhole events such as losses, kicks etc.
– Well Time classification (NB: QC is especially important as these data typically have human entry,
everyone tends to do it differently)
∘ Operational time classification (productive time or non-productive time) and if NPT – accountable
party
∘ Operational time type (e.g. drilling, cementing, casing running, BHA handling etc.)
With all of this data QCed and available at hand, engineers can find relevant wells (even several years
old) that are comparable and compare various aspects important aspects between wells that are crucial to
the success of the well he is planning. Examples being:
– What mud weight should I be using in 12 ¼" hole with 35-degree angle? Check similar wells
with inclination between 30 and 40 deg and find what mud weight was used and any associated
experiences (positive or negative) from these wells.
– What is my estimated pump pressure at the bottom of a challenging section – will my mud
pumps have sufficient effect? Check offset wells with similar hole configuration and check their
pump pressure @ TD
– Am I more likely to come in a situation with liner hanger A than liner hanger B when running?
And if I come into problems, what is the recommended way forward to get back on track? Check
relevant experiences with liner hanger A and B.
– How long is my estimated flat time running production casing? What can be done to minimize
it? Find comparable 5 wellbores, focus in on phase 9 5/8" casing and check what operations time
was spent on. Focus on improving operations that are most time consuming, and definitely focus on
reducing risk for further delays.
Case Study
Figure 1 shows well schematics from two wells of similar character drilled close to each a few years apart,
and Figure 2 shows their corresponding time vs depth curves with darker color denominating NPT. We can
clearly see that drilling portions of the TvD curves for the two wells are more or less parallel (and very
steep), indicating that there's not much benefit to be made by optimizing on-bottom drilling speed. The
exception is the 12 ¼" hole section on the green well (Well A), which has a significantly slower progress
than the blue (Well B). This was due to the fact that for well A, the section was drilled with water based
mud, but for well B, oil based mud was utilized. It should be noted that well A was drilled before well B,
and the lesson was therefore learned not to utilize WBM in this hole section afterwards.
SPE-181027-MS
Figure 1—Well Schematics from 2 reference wells

4
SPE-181027-MS 5
Figure 2—Corresponding time vs depth curves for wells in Figure 1
With the exception of the difference mentioned above, Well A appears to be the most time efficient of
the two, where the main reasons for delay for well B are due to shallow gas (introducing an extra casing
string – 16″ liner) and BOP failure.
Quite obviously, if one was planning to drill a 3rd well in this area with similar design, we would:
1. Make a larger effort into selecting a wellbore location where chance of shallow gas was further reduced
(perhaps predrill a pilot hole?)
2. Utilize OBM in the 12 ¼″ section
3. The BOP failure is something that we would always have to consider as a risk, only measure that
can be taken is to ensure that the rig we have selected as a good long term performance for the BOP.
Thus it should be accepted as a residual risk and it's something we would want to highlight in a
probabilistic approach, perhaps look at several wells drilled with this rig or similar rig with similar
BOP configuration and then estimate possible time delays.
If we take away NPT from the curves shown in Figure 2, we can more clearly see where the differences
in performance are in the wells, this is displayed in Figure 3. The 12 ¼″ drilling still dominates heavily as it
is 28 days for well A and only 7 days for well B. In Figure 4, the drilling of 12 ¼″ section have been taken
away, and thus highlighting the remaining differences between the two wells’ performance.
6 SPE-181027-MS
Figure 3—Modified TvD curves – Productive Time only
Figure 4—Modified TvD curves with drilling of 12 ¼″ hole taken out
In Figure 4, we can clearly see that casing running was much faster on well A than on well B, in fact,
when combining the data from all three casing strings for the two wells we can see that combined time delay
is in excess of 4 days (11 days for well A and 15,2 days for well B).
The follow-up question then becomes – why were flat times significantly longer for well B than for well
A and what can be done to ensure that the background for the good performance from well A in these
particular operations is carried into the new well in the planning stage. For the various operations that the
SPE-181027-MS 7
engineer will be looking at in the planning for the well, this will (or at least should) constitute a large portion
of the work that's required for the well design.
In the case of production casing running, the main cause for the delay for well B was that it was required
to take a wiper trip before POOH to run casing, due to poor hole condition. To take this wiper trip was
probably the right decision as the casing was successfully run to bottom, but it did however take in excess
of 24 hrs. and therefore more than 1 million dollars in rig time alone.
In Figure 5 shown below, the relevant mud parameters for well B are plotted vs casing scheme and
lithology. We can for instance see that the rheology parameters in the 12 ¼″ section (from 2200 m onwards)
vary quite a lot from top to bottom. One should off course be careful to draw the conclusion that this is the
cause for the inadequate hole condition to run casing in, but at least having a schematic like this allows us
browse this data very quickly, even several after the well was drilled. This data can be compared to other
offset wellbores (e.g. Well A) and one can immediately identify differences. Curves that are plotted like this
include drilling parameters, mud parameters and composite log data, available to all in the team.
Figure 5—mud related drilling parameters plotted by depth for Well B.

8 SPE-181027-MS
Conclusions
Data being captured when drilling a well is available in large quantities, but it's often not very accessible.
Instead of looking into the raw data from historic wells afterwards, we tend instead to read through their
(interpreted) final well reports. When wanting answers to our question: "Can we do the job successfully
with the rig / tools we have available?" for a new well, we typically lean up against simulations made in
advanced software where several assumptions often have been made.
It is recommended to change this way of working as one often do not see the full picture from lessons
learned spreadsheets and previously interpreted final well reports – sometimes these have been made in a
hurry while closing out the projects. Instead, have focus on having the raw data highly accessible and on
a platform where one quickly do analyses. The analyses should be entered into the same system, and thus
making the detailed work available to the whole organization – after all it is with the organization one wants
the experience / knowledge, not with the individual team members.
The overall goal is to construct the well safely and as efficiently as possible while meeting the objectives
set ahead of the project. As cost for a well operation is largely time driven, it follows that if you can reduce
the time spent constructing the well, the cost is also reduced by some factor (the correlation between these
two often depends on the contractual setups, day rates or fixed prices).
The focus on having high-quality real-time data available and being able to modify parameters to optimize
progress is good. Monitoring and modifying these parameters also to ensure that downhole events are seen
early and therefore also simpler to mitigate before they escalate gives an even better contribution to well
construction performance. But in our opinion, the possibly largest factors of all for well performance is not
in a high enough degree utilized by drilling & wells teams, this factor being learning from the past.
By implementing a system where you can learn better from the past, the organization may:
1. Build a company internal track record of downhole equipment, therefore reducing risk of picking the
tools with poor historic performance
2. Gain field specific knowledge. What works through these formations and what does not? What should
be done towards end of section to ensure casing running is problem-free. Trajectory planning?
3. Gain insight to what operations on a rig / selection of wells in an area are you spending the most time
on. Can then focus on optimizing these operations to get an overall better well performance.
4. Reduce dependency on advanced simulations
5. Reduce dependency on key personnel with all the experiences / best practices either in a word
document somewhere or not written down.
6. One can utilize a probabilistic approach to well budgeting (both time and cost). Based on historical
well data one can set ranges on expected timings for various operations, NPT amount, risk expectancy
(both likelihood of something occurring and consequence (range) if it occurs)

Not Just Drilling Optimization

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Not Just Drilling Optimization

Uploaded by

Copyright:

Available Formats

SPE-181027-MS

Øystein Andersen, AGR Software

Copyright 2016, Society of Petroleum Engineers

Make the most of historic data

Figure 1—Well Schematics from 2 reference wells

Figure 4—Modified TvD curves with drilling of 12 ¼″ hole taken out

Figure 5—mud related drilling parameters plotted by depth for Well B.

You might also like