Solving Heated Oil Pipeline Problems Via Mixed Integer Nonlinear Programming Approach

Muming Yang, Yakui Huang, Yu‐Hong Dai, Bo LiarXiv 2019 · DOI · arXiv

Abstract

Transporting crude oil efficiently requires balancing the fuel costs of heating with the electricity costs of pumping. This paper formulates the heated oil pipeline problem as a Mixed Integer Nonlinear Programming (MINLP) model. By proving the equivalence between nonconvex and convex continuous relaxations under specific conditions and implementing a preprocessing procedure, the authors developed a branch-and-bound algorithm capable of finding global optima. Enhanced by outer approximation and warm-start techniques, the algorithm achieved a 6.83% reduction in operational costs on a real-world Chinese pipeline.

The Optimization Challenge: Heat vs. Head

Transporting crude oil is a complex thermodynamic and hydraulic challenge. Because crude oil viscosity increases significantly at lower temperatures, it must be heated in furnaces at various stations to maintain flowability. This creates a fundamental economic trade-off:

  1. Heating: Increasing fuel consumption at furnaces reduces oil viscosity.
  2. Pumping: Lower viscosity reduces the "head" (pressure) required from electric pumps to move the oil.
Schematic of a typical pipeline station featuring heating furnaces and pump configurations.

Finding the optimal balance is an MINLP problem. It involves discrete decisions (which pumps to turn on) and nonlinear physics (fluid dynamics and heat loss). Historically, these problems are difficult to solve globally because the mathematical landscape is nonconvex, often leading standard solvers to sub-optimal local minima.

Methodology: From Nonconvex to Convex

The authors bridge the gap between theoretical optimization and practical engineering through three key innovations:

1. Convex Relaxation and Equivalence

The primary theoretical breakthrough is proving that the difficult nonconvex constraints—specifically those related to hydraulic friction—can be treated as convex under certain temperature and pressure bounds. This transformation allows for the use of more robust and efficient global optimization solvers.

2. Preprocessing for Bound Tightening

To ensure the mathematical conditions for convex equivalence are met, the authors introduced a preprocessing routine. This step tightens the bounds on temperature and pressure variables across the pipeline's segments, effectively "cleaning" the search space before the main optimization begins.

Mathematical modeling of pipeline segments between stations, incorporating pressure, temperature, and physical dimensions.

3. Enhanced Branch-and-Bound

A Branch-and-Bound framework manages the discrete pump choices. To accelerate the process, the authors employed:

  • Outer Approximation: Linearizing complex curves to simplify sub-problems.
  • Warm Start Strategy: Allowing the solver to leverage information from previous iterations, reducing computation time for nonlinear sub-problems from 40.8 seconds to 8.6 seconds.

Real-World Validation: The Q-T Pipeline

The model was tested on the Q-T Pipeline in China, a 548.54 km infrastructure with 9 stations. The pipeline was discretized into 131 segments (each ≤ 5 km) to accurately track pressure and temperature gradients across varying terrain and elevation. The study utilized actual viscosity-temperature curves and real-world energy pricing.

Key Results and Insights

  • Significant Cost Savings: The algorithm identified an operational scheme 6.83% cheaper than the existing practical scheme. For large-scale industrial infrastructure, such a margin represents substantial annual financial savings.
  • Global Optimality: Unlike heuristic methods (e.g., genetic algorithms) that offer "good" solutions, this approach provides a theoretical guarantee of finding the absolute best (global) solution.
  • Computational Efficiency: The combination of bound tightening and warm starts makes the algorithm viable for real-time operational planning, solving complex configurations in seconds.

Limitations and Future Directions

While highly effective, the model currently assumes constant friction heat and constant efficiency for pumps and furnaces. Future research could incorporate variable friction heat and nonlinear pump efficiency curves. Additionally, the "equivalence" proof relies on specific friction models; adapting the algorithm for different hydraulic formulas remains an area for further exploration.

Summary updated

Abstract

It is a crucial problem how to heat oil and save running cost for crude oil transport. This paper strictly formulates such a heated oil pipeline problem as a mixed integer nonlinear programming model. Nonconvex and convex continuous relaxations of the model are proposed, which are proved to be equivalent under some suitable conditions. Meanwhile, we provide a preprocessing procedure to guarantee these conditions. Therefore we are able to design a branch-and-bound algorithm for solving the mixed integer nonlinear programming model to global optimality. To make the branch-and-bound algorithm more efficient, an outer approximation method is proposed as well as the technique of warm start is used. The numerical experiments with a real heated oil pipeline problem show that our algorithm achieves a better scheme and can save 6.83% running cost compared with the practical scheme.
BibTeX
@misc{yang2019solving,
  author = {Muming Yang and Yakui Huang and Yu‐Hong Dai and Bo Li},
  title = {{Solving Heated Oil Pipeline Problems Via Mixed Integer Nonlinear Programming Approach}},
  year = {2019},
  eprint = {1907.10812},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1907.10812},
}
Figure