Most rod pump design software runs the deviated wave equation at 50-foot step lengths. For a vertical well with a smooth trajectory, that is probably fine. But for deviated and horizontal wells - the ones that actually need careful design work - a 50-foot step can miss the exact curvature that causes failures.
What Step Length Actually Does
When you run a rod pump simulation, the software interpolates your deviation survey into a continuous well path and then samples it at regular intervals - the step length - to calculate forces along the rod string. A 50-foot step means the model evaluates curvature, side loads, and friction every 50 feet. Anything that happens between those points gets averaged out.
For wells with gradual build rates and minimal tortuosity, that averaging does not change the answer much. But in wells with micro doglegs - small, localized bends caused by frequent changes in inclination and azimuth - a 50-foot step smooths over the peaks that matter most.
The result is a simulation that looks clean but underestimates the forces your rod string will actually see. This is one of the most common issues the PetroBench engineering team encounters in support tickets - an operator runs a design, the model shows comfortable margins, but the well fails repeatedly at the same depth. When we pull the simulation apart, the step length is almost always part of the story.
The Survey Resolution Problem
Step length and survey resolution are connected. A standard MWD survey records inclination and azimuth every 90 to 100 feet during drilling. If your simulation runs at 50-foot steps on 100-foot survey data, you are already interpolating between sparse points. The model cannot capture curvature it was never given.
High-resolution gyro surveys change this picture significantly. These tools record data at 1-foot to 25-foot intervals, revealing trajectory details that drilling surveys miss entirely. In a 2019 presentation at the Southwestern Petroleum Short Course, Norm Hein and Lynn Rowlan showed that as-drilled deviation surveys collected at 100-foot intervals miss inclination and azimuth changes that create micro doglegs, resulting in unexpected tubing failures and shorter run times (Hein & Rowlan, "Dog Leg Severity and Side Load Recommendations," SWPSC 2019). Their recommendation was to obtain a detailed digital deviation survey at 1-foot intervals and then process the resulting data at 30-foot intervals - a significant departure from the industry-standard 100-foot spacing.
The gap between what an MWD survey shows and what a gyro survey reveals can be substantial. Work by Forrester, Boltz, and Shoup published in the Journal of Petroleum Technology found that high-resolution tortuosity logs collected at 1-foot intervals revealed significantly higher tortuosity than MWD data indicated in the same wells (Forrester et al., "High-Resolution Wellbore Tortuosity Logs Improve Completion and Production Equipment Placement," JPT, October 2020). In one Bakken well, standard survey data showed no particular concern, yet the well was failing three times per year at roughly $200,000 in annual workover costs. After running a high-resolution gyro survey and using the results to optimize rod guide placement, failures dropped to zero over the following 12 months.
What Missed Curvature Does to Your Design
When step length is too long or survey resolution too coarse, the simulation underestimates dogleg severity - the normalized measure of wellbore curvature in degrees per 100 feet. Lower predicted DLS means lower predicted side loads, which means the model tells you things are fine in areas where rods are actually buckling or wearing through tubing.
Side loads are lateral forces acting on the rod string where it contacts the tubing wall. They are a function of measured depth, axial loading, buckling tendency, and dogleg severity. In deviated wells, these forces concentrate at couplings and can cause tubing holes, rod breaks, and reduced runtimes if the design does not account for them.
Hein and Rowlan's work proposed moving away from traditional DLS thresholds entirely. The historical guideline - 0 to 3 degrees per 100 feet is acceptable, 3 to 5 causes increased wear, above 5 will cause problems - was developed for vertical and shallow wells. For modern deviated and horizontal wells, they argued that side load is a more useful criterion than DLS alone, recommending that the calculated side load on a 25-foot rod should not exceed 200 pounds (Hein & Rowlan, SWPSC 2019). Their field data showed that wells with side loads above 450 pounds per rod consistently achieved less than four years of runtime. Moreno and Garriz arrived at a complementary finding in their 2019 analysis of sucker rod string dynamics in deviated wells, demonstrating that the wave equation solutions diverge significantly from field-measured loads when the wellbore model fails to capture localized curvature changes - precisely the kind of detail that coarse step lengths smooth away (Moreno & Garriz, Journal of Petroleum Science and Engineering, 2019).
The practical consequence is that a simulation running at 50-foot steps on 100-foot survey data might show acceptable DLS and low side loads in a section where, in reality, micro doglegs are producing side loads well above the 200-pound threshold. The failure shows up in the field, not in the model.
How Step Length Affects Rod Guide Placement
Getting rod guide placement right depends on knowing exactly where side loads are highest. If the simulation smooths over a DLS spike because the step length was too coarse, rod guides end up in the wrong locations.
SPE-190935-MS describes a method for analyzing small-scale tortuosity from high-resolution survey data to calculate the points along the wellbore where the rod string is expected to contact the tubing, enabling more accurate rod guide placement (Forrester et al., "Rod-Guide Placement Based on High-Resolution Tortuosity Analysis of Production Tubing," SPE Artificial Lift Conference, 2018). The approach uses the estimated trajectory of the rod within the production tubing to improve side force calculations - something that is only possible when both the survey resolution and the simulation step length are fine enough to represent the actual wellbore geometry.
In the Bakken case study mentioned earlier, this is exactly what eliminated the three-workovers-per-year problem. The issue was not that rod guides were missing - it was that they were in the wrong places because the design was based on coarse survey data that did not show where the real contact points were.
Practical Recommendations
Not every well needs 10-foot step lengths. The right approach depends on the complexity of the trajectory and the quality of survey data available. Here is a starting framework based on what we see produce reliable results across the wells modeled in PetroBench's simulation engine:
For vertical wells with less than 5 degrees of total inclination, a 50-foot step length is adequate. The trajectory is simple enough that finer resolution does not meaningfully change the force calculations. A well at 8,000 feet TVD with a straight rod string running 86D rods at 8 SPM will produce nearly identical surface dynamometer cards whether you run 50-foot or 25-foot steps. Save the computational budget for wells that need it.
For deviated wells with build rates above 3 degrees per 100 feet or any well with known micro dogleg issues, drop to a 25-foot step length. This is the range Hein and Rowlan recommended for processing gyro survey data, and it captures the curvature features that drive side load spikes. Consider a well in the Permian with a build section from 2,000 to 4,500 feet MD, peak inclination of 45 degrees, running a tapered 76/86D rod string at 6 SPM. At 50-foot steps, the model might show a peak side load of 120 pounds in the build section. Drop to 25-foot steps with matching gyro data and that same section can show localized peaks above 250 pounds - well past the Hein and Rowlan threshold. That difference changes where you place rod guides and whether you need to switch to continuous rod in the build.
For horizontal wells - anything with a lateral section above 80 degrees of inclination - use 10 to 15-foot steps through the curve and lateral. These wells have the most complex force interactions between the rod string and tubing. A Bakken horizontal at 10,200 feet MD with a 6-degree-per-100-foot build rate, 90-degree lateral, and a mixed 76/86/76D rod string running at 5 SPM will show dramatically different rod load distributions at 10-foot versus 50-foot resolution. The lateral section is where rod-on-tubing contact is continuous, and the spacing of rod guides through that section depends entirely on accurately resolving the small trajectory variations that control contact points. Running coarse steps through a lateral is where the most money gets left on the table in terms of preventable failures.
Computational Cost Tradeoffs
The obvious concern with finer step lengths is computational cost. Halving the step length roughly doubles the number of nodes in the simulation, which increases both memory usage and solve time. For a single well, this is trivial on modern hardware - a 10,000-foot well at 10-foot steps produces 1,000 nodes, which any modern solver handles in seconds. The concern becomes real when you are running batch optimizations across hundreds of wells or doing sensitivity analyses that require thousands of simulation passes.
The practical answer is to use variable step lengths along the wellbore. There is no reason to run 10-foot steps through a straight vertical section where nothing interesting is happening. A sensible approach is 50-foot steps through vertical and tangent sections, 25-foot steps through moderate build and drop sections, and 10 to 15-foot steps through the curve and lateral in horizontal wells. PetroBench's simulation engine supports this kind of variable resolution natively - you can assign different step lengths to different measured depth intervals rather than being locked into a single value for the entire wellbore. This is a meaningful difference from tools that force a uniform step length, because it lets you put computational resolution exactly where the physics demands it without paying the cost everywhere else.
We have seen cases in PetroBench where switching from a uniform 50-foot step to a variable resolution scheme changed the predicted peak polished rod load by 8 to 12 percent and shifted the predicted failure location by several hundred feet. Those are not rounding errors. They are the difference between a rod string that makes it to the next workover cycle and one that fails at month four.
The key is matching step length to survey resolution. Running a 10-foot step on 100-foot survey data does not add real accuracy - it just interpolates more finely between the same sparse points. The value comes from pairing high-resolution survey data with a step length short enough to use it.
For deviated wells where failures are occurring in locations the model did not predict, the first thing to check is whether the simulation is actually seeing the wellbore geometry. A gyro survey processed at 25 to 30-foot intervals, combined with a matching step length in the simulation, will often reveal the curvature that explains the failures.
References
1. Hein, N. and Rowlan, L. "Dog Leg Severity (DLS) and Side Load (SL) Recommendations to Drilling." Southwestern Petroleum Short Course, 2019.
2. Forrester, S., Boltz, J., and Shoup, R. "High-Resolution Wellbore Tortuosity Logs Improve Completion and Production Equipment Placement." Journal of Petroleum Technology, October 2020.
3. Forrester, S. et al. "Rod-Guide Placement Based on High-Resolution Tortuosity Analysis of Production Tubing." SPE-190935-MS, SPE Artificial Lift Conference and Exhibition, 2018.
4. Moreno, G. and Garriz, A. "Sucker Rod String Dynamics in Deviated Wells." Journal of Petroleum Science and Engineering, 2019.
