Water Values in Hydropower Planning

Water values (norsk: vannverdier) are the single most important concept in hydropower production planning. This document explains what they are, why we need them, and how they fit into the planning hierarchy.


The core problem

A hydropower reservoir is a battery. Water stored today can be converted to electricity tomorrow — or next month — or next year. The key question is:

When is the right time to release water through the turbines?

If you release too early (e.g. in spring when prices are low), you waste potential revenue from autumn/winter when prices are high. If you release too late, the reservoir may overflow in spring and water is spilled — completely wasted.

This is the intertemporal trade-off at the heart of hydropower planning.


What is a water value?

The water value at a given week and reservoir level is the marginal value of one additional unit of water in storage:

WV(t, v) = ∂V(t, v) / ∂v   [NOK/Mm³]

where V(t, v) is the maximum expected future revenue from week t onward, given reservoir volume v.

Intuitively: if the water value at week 30 with volume 60 Mm³ is 500 NOK/Mm³, then having one more Mm³ of water in the reservoir at that point is worth 500 NOK more in expected future revenue.

Water value as a shadow price

Water value is the shadow price of the reservoir storage constraint in a long-term optimisation model. It answers the question:

"If I relaxed the storage constraint by 1 Mm³, how much more revenue could I earn?"

This interpretation connects water values directly to linear programming duality (see Kompendium 1, §5.3).


Why do water values depend on week and volume?

Seasonal variation (week)

Norwegian reservoirs typically:

  • Fill up in spring/summer (snowmelt and rainfall)
  • Are drawn down in autumn/winter (high demand, high prices)

The marginal value of water changes through the year:

  • Spring: water value is low — inflow is arriving soon, reservoir fills automatically
  • Late autumn: water value is high — no inflow for months, must conserve to cover winter demand

Volume dependence

The water value also depends on how full the reservoir is:

  • Low reservoir → water value is high — scarce resource, use carefully
  • High reservoir → water value is lower — risk of spill, better to release

This is why we compute WV(t, v) — a 2D surface, not a single number.


The water value table

The output of the SDP computation is a discrete 2D table:

WV[t, i]  →  NOK/Mm³

where:
  t  = week (1–52)
  i  = volume grid index (1 = v_min, N = v_max)

This table is precomputed once (typically annually) and used as:

  1. End condition for short-term LP scheduling (Phase 2)
  2. Operating policy in simulation (Phase 3)
  3. Bid curve basis for spot market participation

Connection to the planning hierarchy

Long-term SDP (5 years)
       │
       │  produces
       ▼
  Water Value Tables  ← ← ← ← ← ← ← ← ← updated annually
       │
       │  used as end-condition in
       ▼
  Medium-term LP (1 season)
       │
       │  produces
       ▼
  Weekly schedules + refined water values
       │
       │  used in
       ▼
  Short-term LP/MIP (1–2 weeks)

The water value table is the interface between long-term and short-term planning — it encodes the future value of water at the boundary between horizons.


Key references

  • Kompendium 1, Chapter 6: Vannverdiberegninger
  • Kompendium 1, §6.1: Single-reservoir deterministic DP
  • Kompendium 1, §6.2: Vansimtap — the Norwegian industry SDP tool
  • Kompendium 1, §6.3: Stochastic inflow representation