Stick-Slip Index (SSI): Friction & Drilling

• Stick-Slip Index (SSI) is a quantitative metric that captures the onset, severity, and dynamics of stick-slip transitions in both tribological sliding and mechanical drilling.
• It incorporates effective energy barriers, aggregate stiffness, and periodic length scales from the Prandtl–Tomlinson model to classify frictional regimes.
• In drilling, SSI evaluates torsional vibrations from downhole bit speed data, enabling automated detection of severe mechanical instability.

The Stick-Slip Index (SSI) is a domain-specific quantitative metric that captures the onset, severity, and dynamical characteristics of stick-slip transitions in both tribological sliding interfaces and mechanical drilling applications. In tribology, it serves as a predictive dimensionless criterion for frictional regime transitions in structurally lubric 2D contacts, including graphene-like systems. In drilling, SSI is the established scalar measure for torsional downhole vibrations, enabling automated severity assessment from monitored rotational velocities. The concept unifies mechanical instability principles from the Prandtl–Tomlinson (PT) model with normalized outcome-based detection for coupled dynamical systems.

1. Mathematical Formulation Across Domains

Structurally Lubric 2D Interfaces

For 2D layer sliding, the Stick-Slip Index $\eta_\mathrm{eff}$ is rigorously defined by

$$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$

where $U_\mathrm{eff}$ denotes the effective energy barrier along the sliding direction, $K_\mathrm{eff}$ the aggregate stiffness (combining pulling mechanism and layer elasticity), and $a_\mathrm{eff}$ the effective periodicity, which is within 10% of the substrate lattice constant $a$ for stiff materials. This PT-type criterion directly classifies sliding regimes:

• $\eta_\mathrm{eff} < 1$: smooth, structurally lubric, velocity-linear sliding
• $\eta_\mathrm{eff} > 1$: mechanical instability, stick-slip cycles, increased force noise

Drilling Applications

In well drilling, the scalar Stick-Slip Index SSI is calculated from a 60-second window of downhole bit speed $\{\omega_\mathrm{Bit}(t)\}$ sampled at 1 Hz:

$$\mathrm{SSI} = \frac{\max_t\,\omega_\mathrm{Bit}(t) - \min_t\,\omega_\mathrm{Bit}(t)}{\overline{\omega_\mathrm{Bit}}}$$

where $$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$0. Values close to zero represent negligible stick-slip, and $$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$1 designates severe cyclic events (Yahia et al., 6 Jan 2026).

2. Physical Origin and Regime Classification

The SSI for tribological systems derives from mechanical instability in a periodic potential. It generalizes the 1D PT instability criterion $$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$2 to collective island or flake motion, incorporating lattice, elastic, and energetic considerations (Wang et al., 2024). For drilling, SSI reflects transient deviations in bit angular speed, with high SSI diagnosing destructive torsional vibrations.

Stick-slip transitions coincide with:

• Exceeding $$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$3 in sliding islands, triggering discrete instability events
• Elevated SSI in drilling data, indicating amplitude spikes in rotational speed

3. Computation of Effective Parameters

Tribological SSI ($$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$4)

- $$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$5 (Energy Barrier)

$$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$6, assessed along sliding direction. Originates from uncompensated edge atom contributions and scales gently: $$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$7 for circular islands of area $$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$8 (diameter $$\eta_\mathrm{eff} = \frac{2\pi^2\, U_\mathrm{eff}}{K_\mathrm{eff}\, a_\mathrm{eff}^2}$$9), since edge moiré node count increases $U_\mathrm{eff}$0.

- $U_\mathrm{eff}$1 (Aggregate Stiffness)

$U_\mathrm{eff}$2 harmonizes external pulling stiffness $U_\mathrm{eff}$3 (e.g., AFM cantilever) with island internal stiffness $U_\mathrm{eff}$4 for Young’s modulus $U_\mathrm{eff}$5, effective thickness $U_\mathrm{eff}$6, and island size $U_\mathrm{eff}$7:

$U_\mathrm{eff}$8

At nanoscale, typically $U_\mathrm{eff}$9.

- $K_\mathrm{eff}$0 (Periodic Length Scale)

$K_\mathrm{eff}$1 is linked to mechanical instability displacement, but for graphene/rigid substrates, $K_\mathrm{eff}$2 holds within 10% error.

Drilling SSI (Time-Series)

SSI computation requires synchronized sensing:

• Surface variables: $K_\mathrm{eff}$3, WOB$K_\mathrm{eff}$4, ROP$K_\mathrm{eff}$5, $K_\mathrm{eff}$6, RPM$K_\mathrm{eff}$7
• Interpolation of bit speeds and time windowing between surface and downhole streams
• Use of non-overlapping 60 s data blocks

4. Model Architectures and Algorithmic Approaches

Drilling SSI Prediction

Three regression approaches have been benchmarked (Yahia et al., 6 Jan 2026):

• Baseline LSTM: 6-layer stack (64 units/layer, LN), linear output, trained with empirical MSE minimization
• Adversarial Domain Generalization (ADG): Generator, SSI-predictor, and domain classifier (gradient reversal). Min-max objective aligns domain-invariant features, governed by hyperparameter $K_\mathrm{eff}$8
• Invariant Risk Minimization (IRM): Same generator/predictor, IRM penalty via domain-wise gradients at $K_\mathrm{eff}$9, hyperparameter $a_\mathrm{eff}$0

Hyperparameters are optimized via grid search on held-out well partitions, with chosen values: regularization $a_\mathrm{eff}$1, LSTM depth=6, $a_\mathrm{eff}$2, $a_\mathrm{eff}$3.

Performance Metrics:

• Primary: normalized dynamic time warping (DTW) of SSI, averaged over 5 runs
• Severe event recall: Baseline $a_\mathrm{eff}$420%, ADG/IRM $a_\mathrm{eff}$560%

Well	Baseline DTW	IRM DTW	ADG DTW
7	0.136	0.120	0.122
8	0.086	0.080	0.075
9	0.107	0.100	0.097

This suggests adversarial domain alignment yields the greatest cross-well generalization. Transfer learning further improves results after fine-tuning with partial labeled sequences.

5. Application Domains and Operational Impact

Frictional Interfaces

$a_\mathrm{eff}$6 predicts sliding regime across arbitrary size, twist, direction, or defect state in structurally lubric 2D contacts. It enables a priori determination of whether extended islands will display smooth sliding or discrete stick-slip, facilitating device and materials design involving superlubricity (Wang et al., 2024).

Drilling Operations

SSI enables automated, surface-data-based detection of severe torsional vibrations, reducing the need for downhole sensor deployments. Severe event recall using advanced domain generalization models jumps from 20% to 60%, directly impacting drilling reliability and maintenance (Yahia et al., 6 Jan 2026).

A plausible implication is that robust surface-log SSI prediction may permit real-time intervention and preempt catastrophic bit failure.

6. Practical Computation Procedures

For $a_\mathrm{eff}$7:

1. Extract energy profile $a_\mathrm{eff}$8 from simulation or experiment
2. Compute $a_\mathrm{eff}$9
3. Approximate $a$0 by substrate constant $a$1
4. Estimate $a$2 and, if needed, $a$3; combine to $a$4
5. Form $a$5
6. Predict regime: $a$6 (smooth); $a$7 (stick-slip)

For drilling SSI:

1. Segment 60 s intervals of surface and bit-speed data
2. Compute SSI per window as specified
3. Apply trained regression model for real-time event detection

7. Limitations, Model Misfit, and Future Directions

Common misprediction causes in drilling SSI include synchronization error (surface vs. downhole), deep attenuation of stick-slip (surface-invisible), mislabeling of peaks, and domain mismatch. For tribological $a$8, estimation depends on energy landscape extraction precision and accurate stiffness calibration.

Recommended future directions (Yahia et al., 6 Jan 2026):

• Train on larger, more diverse well sets
• Explore adversarial domain adaptation with few labels
• Incorporate higher-frequency measurements and physics-informed inductive biases
• Formal statistical validation of performance gains across algorithms

For structurally lubric interfaces, further validation of $a$9 scaling and exploration of multi-defect, multi-twist systems are warranted.

• "Effective stick-slip parameter for structurally lubric 2D interface friction" (Wang et al., 2024)
• "Domain Generalization for Time Series: Enhancing Drilling Regression Models for Stick-Slip Index Prediction" (Yahia et al., 6 Jan 2026)