Every Drop Counts — Is It Possible?

A significant challenge during development of the 42.5 MW Sibulan A and B hydro plants in the Philippines was to use every drop of water coming down the river to generate power. Owner Hedcor Inc. shows how results of hydraulic studies were used to maximize revenue from the two facilities.

By Rafael B. Macabiog, Augusto O. Rodero, Cherry L. Malicay and Julius Peterson C. Barcena

This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.

The Sibulan A and B hydro plants were built to provide daily peaking operation for owner Hedcor Inc. The two cascading plants harness the Sibulan and Baroring rivers in Mindanao, Philippines, and have a total capacity of 42.5 MW and expected annual output of 216.4 GWh.

The climate in this region consists of rainfall distributed almost uniformly throughout the year. The unique challenge in developing and operating these two projects is to attain generation and revenue that matches the forecast from when these plants were conceptualized. Both operate as daily peaking plants, making use of their total combined pondage of 150,000 m3 to store water, and they must be able to attain the expected revenue using time of use rates, which vary hourly. With these constraints, every drop of flow counts, and to make this possible is the challenge Hedcor faced.

To support operations in attaining the expected generation and revenue, Hedcor performed a study of the hydrological approach needed. The optimization model used considered three different operational arrangements: revenue-based operations, three-peak operation and load requirements of the power distribution utility. The three-peak optimization was dropped because it did not prove to benefit the plants. The plants had been operating based on load requirements, but plans are in place to convert to revenue-based operations.

The project has led to a new development scheme for Hedcor with regard to its future hydropower projects. It also has opened the opportunity of getting higher revenues by taking advantage of the varying hourly energy rates.

The Sibulan A and B projects

Construction of the two run-of-river Sibulan plants was undertaken by Hedcor Sibulan Inc. (HSI), a special-purpose company organized to enter into contracts for plant development, construction and operation. Power from the plants is sold to the city of Davao through Davao Light and Power Company (DLPC), the third largest power distribution utility in the country. In March 2007, Hedcor signed a Power Sales Agreement (PSA) with DLPC after winning an open bid to supply DLPC with 200 GWh of energy annually.

Ground was broken for both projects in June 2007. The Energy Regulatory Commission (ERC) approved the PSA in April 2008. And in May 2008, HSI obtained a PHP3.5 billion (US$83.6 million) syndicated loan from a consortium of local banks to fund development of these projects.

The Sibulan project was registered under the United Nations Framework Convention on Climate Change (UNFCC) as a Clean Development Mechanism (CDM) project in June 2008. It is the first CDM-registered hydro project in the Philippines and the largest among all registered renewable energy projects in terms of generating capacity. Through the CDM, HSI earns and sells “certified emission reduction” units as a result of reducing greenhouse gas emissions by more than 95,000 tonnes of CO2 equivalents annually.

ERC awarded the Certificate of Compliance for 26 MW Sibulan B in May 2009, while the certificate for 16.5 MW Sibulan A was awarded in August 2009. Both plants were inaugurated on September 16, 2010, and have been operating successfully since that date.

Operating as daily peaking plants

Sibulan A has an impounding basin of 100,000 m3 and Sibulan B 50,000 m3. The basins can be used to store water during times of low river flow and release it as needed to attain the desired load. The Sibulan A basin also serves as additional storage for Sibulan B. This storage is required almost 80% of the days of an average year. The stored water is to be released when time of use rates are high. These rates vary hourly, and water release also needs to conform to the load nomination criteria established as follows:

— Revenue based on time of use rates;

— Three-peak demand; and

— DLPC load requirement.

The Sibulan A and B plants each contain two Pelton units that operate at 365.75 metres and 307.61 metres of head with design discharges of 5.23 m3/sec and 9.89 m3/sec, respectively. The intakes are run-of-river type, with Sibulan A having two main intakes and Sibulan B having one main intake and two tributary intakes.

The plants operate to allow full load to run over a certain duration of the day. The duration and scheduling both depend on the available river flow and load nomination criteria.

Hydrological challenges of daily peaking operation

There are several hydrological challenges that need to be addressed in operating the Sibulan plants for daily peaking.

The hydrology study of the plants during the design stage simulated average daily flows. These flows were input into the generation model to calculate the plants’ generation and revenue. The model simulated two outputs: run-of-river and daily peaking operation. Results showed daily peaking operation provides higher revenue. The difference in 2008 was PHP25 million ($597,600) annually.

The daily peaking gross annual forecast for the two Sibulan plants was set at 216 GWh. This is an average gross annual forecast calculated from 19 years of simulated daily flows. Generation varies annually within those 19 years, ranging from 129 GWh to 308 GWh annually. From 2008 to 2010, river flows were recorded at the intake locations, and the corresponding generation for the flows in 2008, 2009 and 2010 were 257 GWh, 244 GWh and 181 GWh. In 2010, annual rainfall was below normal, and this is reflected in the results of the simulated generation. Beginning in 2011, the monitoring for monthly to annual generation was based on the actual generation of the plants, instead of the forecasted generation of 216 GWh. River flows continue to be measured as a reference to validate the hydrologic study.

In addition to the forecast for monthly to annual generation, it is necessary to forecast for the day’s load, hours ahead of time. This is the hydrological challenge the plants face in operation. DLPC requires that HSI provide a day-ahead forecast for each plant, supplied at 3 p.m. for the 24 hours of flows of the following day, starting at 1 a.m.

The forecasted versus actual hourly flows give a disparity between the nominated loads and actual loads. As a result, the target of +/-3% difference between the day-ahead forecast and actual load is difficult to attain without an effective and reliable inflow forecasting model.

An additional concern is the ability of operations personnel to manage the available river flow and to store/release water at the best time.

In the design stage, flows were just averaged for the day while in the operation stage — flows now vary hourly. Any time of the day, it could rain, and that will change the river flow drastically.

Approach to Sibulan’s hydrological challenges

For any given day, a specific volume of water flows along the river. The rule is to utilize it fully without having any unnecessary spill at the weir. Whenever river flow is below the design discharge, there must be no spill other than the required compensation flow. Compensation flow is 10% of the river’s dependable flow, which is 80% exceedance from the annual flow duration curve. This is about 260 litres/sec for Sibulan A and 270 litres/sec for Sibulan B.

To approach Sibulan’s hydrological challenges, three tasks were performed:

— Set up HydroMet system;

— Establish an hourly inflow forecasting model; and

— Develop an optimization model.

Setting up HydroMet

The HydroMet system refers to the collection and real-time transmission of rainfall and stream flow data. The HydroMet concept of installing automatic hydrologic stations to provide real-time information is widely recognized. It aims to adopt existing technology, particularly for the collection and transmission of data.

Table 1 Stream Flow and Rainfall Stations Installed

There are five stream flow stations installed in the catchment area of the two Sibulan plants. Table 1 on page 46 provides the location, status, observation dates, type of instruments used and frequency of data collection of the stream flow and rainfall stations that were installed at the Sibulan A and B plants. The data collected is transmitted real time to the Sibulan A and B powerhouses using a combination of Global System for Mobile Communications (GSM), radio frequency and fiber optics.

The hydrologic information is used to forecast hourly inflows and for the succeeding hours after the time the forecast was made.

Establishing an hourly inflow forecasting model

Forecasting is defined as the estimation of conditions at a specific future time or during a specific time interval. Thus, to determine an operating policy that maximizes Sibulan’s hydropower output, a good forecast of hourly inflows is essential.

For the day-ahead forecast, the autoregressive moving average (ARMA) model, a popular streamflow forecasting tool, was used.

The results of the ARMA model are supplied at 3 p.m. for the 24-hour flows of the following day starting at 1 a.m. The difference between the forecasted versus actual hourly flows of the same day indicate that forecasting flows as early as 3 p.m. for the hourly flows for the following day starting at 1 a.m. is not realistic, with little chance of attaining +/-3% difference.

Figure 1 Model Result of Three to Four Hours Lead Time

The ARMA model was used for short lead times — three to four hours ahead. Figures 1 and 2, show the results and indicate that forecasted flows for the one hour lead time are closer to the actual flows. The hourly inflow forecast model derived using the ARMA model can be effective to guide operation, forecasting loads three to four hours ahead of time. The forecast is updated every hour for any changes in flow pattern.

Figure 2 ARMA Model Result at One Hour Lead Time

Develop an optimization model

The optimization model is a program that identifies optimal range of loads for a policy that will cover operation of the Sibulan plants. The objective is to optimize revenue while satisfying the generation demand. The decisions that need to be made include:

— Volume of water to release;

— Particular time to release this particular volume of water; and

— What volume to distribute in what time frame.

The system constraints to this decision are from the fixed layout of the hydro plants (intake, impounding basin, plant capacities, etc) and from hydraulic laws and principles. Economic parameters include time of use rates and load generation demand.

The optimization models developed follow the below load nomination criteria:

— Revenue based on time of use. The model maximizes daily plant revenues through designating higher volume outflows to hours with higher time of use rates and lesser volume outflows to hours with lower time of use rates. The filling (store water) of the impounding basis is scheduled when the time of use rates are low. Figure 3 shows a sample of a day of operation for the plants.

Figure 3 Operation Based on Time of Use Revenue

— Three-peak demand. The model satisfies the supply daily generation demand by allocating higher volume outflows to peak hours, such as 9 a.m.- 11 a.m., 1 p.m.-3 p.m., and 6 p.m.- 9 p.m.). A minimum of 10 MW was allocated at filling time (11 a.m.-1 p.m. and 3 p.m.-6 p.m.). This operation scenario was set due to preidentified three-peak areas that were coordinated with DLPC. Figure 4, shows a sample day of operation for the plants.

Figure 4 Operation Based on Three Peak Demand

— DLPC load demand requirement. Designate higher volume outflows to hours with higher demand rates and lesser volume outflows to hours with lower demand rates. It is based on DLPC’s demand curve. Figure 5, shows a sample day of operation for the plants.

Figure 5 Operation Based on Customer Load Demand

Among the three optimization models, the revenue based on the time of use model was higher as compared to the three-peak demand and DLPC load demand requirement models. Revenue in pesos was 2,959,358 using the revenue optimization model, 2,870,346 using the three-peak demand model and 2,869,259 using the DLPC demand optimization model.

Note that there are other factors besides revenue that need to be considered in making the final decision to nominate the load. Within the day of operation, DLPC requests hourly loads that are not in the day-ahead nomination (generation plan) they issued the day before. This is during instances when DLPC receives an unexpected power demand from the city or when problems arise with the other generating facilities that provide power to DLPC.

Every flow counts, is it possible?

The hydrology study of the Sibulan A and B plants contained simulated flows that were used to forecast the expected annual generation at these plants and became the reference of operation to monitor annual generation targets. To achieve this goal, every flow counts. Every hour of every day, the river flow varies, and the objective is to fully utilize these flows for energy. The hydrological challenges behind the operation of the Sibulan facilities as daily peaking plants were as follows:

— Provide daily average flows to DLPC, which was found to be above the +/-3% target when done at 3 p.m. for the 24 hour flows of the following day starting at 1 a.m.; and

— Establish an hourly inflow forecasting model that can be effective to guide operation, nominating loads three to four hours ahead of time. Likewise, the forecast is updated every hour for any changes in flow pattern.

The approach to these hydrological challenges, to make every flow count, was to set-up the HydroMet System, establish an hourly inflow forecasting model, and develop an optimization model. All three tasks are linked together in that the success of the whole depends on each part.

It is the desire of HSI to continue monitoring and recording the advantages gained as operation of these two plants moves forward.

The installation of the telemetry system was suspended because of the intermittent network signals, thereby there is no assurance for real-time data hourly. This is due mainly to the remote location of three of the intakes. Subsequently, Hedcor is studying other alternatives to have the HydroMet station’s telemetry working, as this is the only component of the project remaining needed to fully serve the intended purpose. These include using LAN or WIFI signals to transmit data, real-time. While working on this, the forecasting and optimization models developed also are continuously being calibrated.

Rafael Macabiog is assistant vice president of engineering and construction; Augusto Rodero is supervisor and Cherry Malicay is senior engineer of the water resources engineering and management team for the engineering and construction department; and Julius Peterson Barcena is a water resources engineer with Hedcor Inc. in the Philippines. Hedcor is a wholly-owned subsidiary of Aboitiz Power Corp.

Previous articleWhy Every Solar Company Should Have a Blog
Next articleEco-friendly Farming: Sowing the Seeds of Renewable Energy

No posts to display