ABSTRACT

This paper presents a numerical test problem that was used as the basis for ranking the river basin management models that participated in a tender for a World Bank-funded project in India. The problem is deterministic, and it involves a set of constraints and management objectives that were of interest to the Narmada Control Authority. The test problem had five reservoirs, seven off-stream and four on-stream water demands, steady-state inflows based on a 10-daily time step over a period of nine years, and relatively simple reservoir operating rules. The paper describes the problem and discusses the solutions that were obtained from various vendors. The principal finding is that simulation models should not be used to solve optimization problems. The paper also highlights the importance of establishing benchmarks as an important tool for model evaluation. Input data and the current benchmark solution are available for download.

摘要

本文提出了一个数值测试问题，该问题被用作对参与世界银行资助的印度项目招标的流域管理模型进行排名的基础。这个问题是确定性的，它涉及到 Narmada 控制局感兴趣的一组约束和管理目标。测试问题有 5 个水库、7 个非流水需求和 4 个流水需水量、基于 10 天时间步长在 9 年内的稳态流入量以及相对简单的水库运行规则。本文描述了该问题并讨论了从不同供应商处获得的解决方案。主要发现是不应使用仿真模型来解决优化问题。该文件还强调了建立基准作为模型评估的重要工具的重要性。输入数据和当前基准解决方案可供下载。