A 100-foot survey and a 10-foot survey on the same well predict different failure locations. Not slightly different - structurally different. The 100-foot survey smooths over micro doglegs that do not average out. In a deviated well, those micro doglegs are where the rod string actually contacts the tubing, where side loads concentrate, and where the next failure is going to happen. If the survey that fed your simulation missed them, the simulation is confident about the wrong spots.
What the Standard Survey Misses
Standard drilling surveys typically capture inclination and azimuth every 95 to 100 feet. That interval was designed for drilling navigation, not rod string analysis. Between those points, the wellbore does not follow a clean arc. It wobbles. It has slider-rotary patterns, borehole spiraling from vertical drilling, localized bends from bit walk and formation heterogeneity. These micro doglegs - SPE papers have documented them for decades - are real curvature that affects real rod behavior.
Gyroscopic surveys run at 5 to 25 foot intervals catch these. DLS values that looked reasonable at 100-foot resolution often show spikes of hundreds of degrees per 100 feet higher when you actually measure the path the wellbore takes. Side load predictions that scale with DLS and axial tension shift by hundreds of pounds in critical sections.
Step Length Multiplies the Error
Once you have a survey, the wave equation samples it at a step length you choose. A 50-foot step is the common default, inherited from when rod design tools could not handle anything finer. Even if you have 10-foot gyro data, a 50-foot calculation step throws away four out of every five data points and smooths the trajectory in the process. The simulation runs on the smoothed version, not the actual well.
Match step length to survey resolution. If you have 10-foot survey data, use a 10-foot step. The only reason not to is computational cost, and that constraint is mostly a legacy of desktop software that could not handle the data volume.
What Step Length to Use
Vertical or simple trajectory wells: 50 to 100 feet. Low curvature variability, minimal error even at coarser steps.
Moderately deviated wells: 25 to 50 feet. Captures moderate doglegs and shifts predicted failure points into alignment with field observations.
Highly deviated or horizontal wells: 10 to 25 feet. Essential for capturing micro doglegs and Coulomb friction effects. At 50 feet in these wells, you are modeling a different trajectory than the one the rod is actually running through.
What the Errors Look Like in the Field
Undetected micro doglegs underestimate DLS and side loads. The simulation predicts no high side loading on a particular joint, yet that joint is where the string fails - because the actual curvature concentrated side load there and the simulation did not see it. The fix is not better rod guides or different rod grades. It is a better model of the well.
Missed curvature also minimizes Coulomb friction in dynamic models. Friction-induced behaviors like micro-sticking alter downhole card shape and real rod behavior. Without high-resolution input, dynamic simulations under-predict these effects, and the card you compare to the field does not match.
Practical Path Forward
Run gyro surveys routinely on deviated or horizontal wells. Feed the full survey into simulation without decimation. Use a step length that matches survey resolution, not the historical 50-foot default. Validate the model against actual failure locations from prior workovers - the field data tells you whether your simulation is seeing the right wellbore.
Bottom Line
Survey resolution and step length are not tuning parameters. They define which well you are simulating. In deviated wells, running with legacy defaults can predict confidently about the wrong places on the string. High-resolution survey data and matching step lengths fix the input problem, and everything downstream - rod guide placement, material selection, run life predictions - gets more accurate because the input does.