Lecture 2 Reservoir Capacity Yield Analysis (how to size a reservoir and measure its performance) CVEN 5838 Aug 28, 2008 Reservoir Yield: Controlled Release from a reservoir (or system of reservoirs). Often expressed as a ratio of % of mean annual flow. E.g., 70% yield means the reservoir can provide a regulated release of 70% of the mean annual flow. The Yield depends on the active storage capacity of the reservoir Reliability of Yield: probability that a reservoir will be able to meet the demand in any particular time interval (usually a year) Reliability = Ns/N Ns is number of intervals in which demand was met; N is the total

number of intervals Firm Yield: can be met 100% of time CVEN 5838 Aug 28, 2008 Estimating Active Storage Capacity Mass Balance Equation of Reservoirs St-1 + Qt Rt Lt = St Where St-1 is storage at end of previous time interval St is storage at end of current time interval Qt is inflows at current time interval Rt is release at current time interval Lt is loss (evap/seepage) at current time interval Reservoirs have a fixed storage capacity, K, so St <= K for each interval CVEN 5838 Aug 28, 2008 Mass Diagram Analysis (Rippl) Method

Find the maximum positive cumulative difference between a sequence of prespecified (desired) reservoir releases Rt and known inflows Qt. Record of historical inflows is used, typically 40 Example: Nine period-of-record flows: 35 [1, 3, 3, 5, 8, 6, 7, 2, 1] 2. 3. Plot cumulatives Add demand line Find max deficit inflow volume

1. 30 25 Rt=4 20 15 10 5 0 1 CVEN 5838 Aug 28, 2008 2

3 Inflow Cumulative Inflow 4 5 time 6 7 8 9 Mass Diagram Analysis (Rippl) Method

Repeat the hydrologic sequence in case critical period is at end Start demand line at full reservoir Calculate Active Storage Ka as max accumulated storage over 2 successive periods of record CVEN 5838 Aug 28, 2008 Rt=4 Mass Diagram Analysis (Rippl) Method The maximum capacity is determined by the Critical Period Critical Drawdown Period: Longest period from full reservoir condition to emptiness Critical Period:

Longest period from full condition, through emptiness and to a full condition again CVEN 5838 Aug 28, 2008 Problems with Mass Diagram Method Reservoir release needs to be constant (this is not accurate for monthly interval as demands are often seasonal) Assumes that future hydrology is like the past

Cannot compute storage size for a given reliability Evaporation and other losses that depend on level in reservoir cannot be factored into analysis CVEN 5838 Aug 28, 2008 Reservoir Planning by Simulation Simulation models are based on mass balance equations and take into account all time-varying processes as well as elevation-based processes (losses) Simulation based techniques are called Behavior Analysis

Why is it difficult to use simulation for reservoir sizing? Another type of simulation is based on deficit instead of storage. This solves some of the simulation problems. Approach is called Sequent Peak Algorithm CVEN 5838 Aug 28, 2008 Sequent Peak Analysis Let Kt be the maximum total storage requirements needed; Set K 0 = 0 Example: Nine period-of-record flows: [1, 3, 3, 5, 8, 6, 7, 2, 1] Apply the equation consecutively for up to 2x the total length of

the record. The maximum of all Kt is the required storage capacity for the specified releases Rt and inflows Qt CVEN 5838 Aug 28, 2008 Accounting for Losses Evaporation and Rainfall (net losses) depend on Surface Area Surface area depends on storage non linearly: But can be approximated linearly: CVEN 5838 Aug 28, 2008 Accounting for losses

Losses can alter the critical period and change the storage capacity Net evap (equations) can be included in storage based simulations directly Lele and Montaseri developed methods for including evap in SPA in an interative approach (See McMahon et al., Sequent Peak Algorithm paper on class website) An alternative evap model can be used directly with SPA

CVEN 5838 Aug 28, 2008 Reliability Reliability (time-based) - Ns/N Resiliance Speed of recovery Vulnerability severity of failure (volumetric reliability) (Tradeoffs on these 3 lead to very tricky design; several authors have developed indices that attempt to combine these (Loucks, 1997; Zongxue, 1998) Reliability and Vulnerability can be integrated into SPA: Determine number of failure periods permitted (from reliability) Determine reduced maximum reduced capacity permitted (from vulnerability)

See example in McMahon CVEN 5838 Aug 28, 2008 Classification of S-Y-P Procedures (Storage Yield Performance) CVEN 5838 Aug 28, 2008 Behavior Analysis storage based simulation (detailed simulations of processes) Case 1: Estimate System Performance Active capacity of reservoir is known. Series of demands is given (could be stochastic). Starting condition is known or assumed. Historical or synthetic inflows are used (could be many traces) (index sequential or monte carlo) Case 2: Find size of reservoir under consideration Series of demands is known (could be stochastic)

series of inflows is given (historical or generated stochastic) Performance criterion is specified Estimate reservoir capacity; do simulations; adjust until get performance desired Case 3: Estimate Yield of an existing reservoir Specify performance criteria (reliability, vulnerability, maybe resiliance) Use historic or stochastic hydrology Estimate demand (yield) and iterate until meet performance criteria CVEN 5838 Aug 28, 2008 Need to know and model detailed operating procedures Additional important topics Optimization techniques simple ones can give identical results to SPA more complex methods can include cost functions Multiple reservoir systems single-reservoir techniques can be extended to multireservoir systems

determine size at each site specify operating policies for releases and demands (heuristic policies have been developed) typical objective is to keep reservoirs balanced CVEN 5838 Aug 28, 2008