Managing Sediment in Peru’s Pativilca River Basin

After a high concentration of suspended solids in Peru’s Pativilca River began causing damage and downtime at the 43 MW Cahua hydro plant, operator Statkraft Peru undertook an upgrade process that includes improved runners and development of new operating protocol.

By Anibal Maita Espinoza

The Pativilca River, located in the Pacific watershed in Peru, carries large amount of suspended sediment in times of flood and the 43 MW Cahua hydropower plant is located on this river. The river carries around 2.4 million tons of sediment per year – 2.16 million tons of which are carried during the flood season that runs January through April.

It was necessary to stop operating the Cahua plant when suspended solids in the river reached 3 grams per liter (g/l), per the threshold established by the runners’ original design specifications. This situation resulted in fewer working hours for the two turbine-generator units, higher operating costs and lower income from generation.

The accumulated periods of plant downtime due to the presence of solids ranged from 15 to 30 days per year, equating to a loss of 19.2 GWh lost annually.

The Cahua hydropower plant

The run-of-river Cahua project is owned by Statkraft Peru and located 60 km upstream from the mouth of the Pativilca River. The plant, completed in 1967 with a pair of Riva-manufactured Francis turbines, is designed for a flow of 24 m3 per second and a gross head of 215 m.

The intake includes a mixed barrage – the mobile part of which is composed of two radial gates located on the intake side and the fixed part composed of a weir operating under open flow conditions.

The desander – present since the plant opened – is composed of eight bays working in tandem with water conveyed through an open-flow tunnel 7.88 km long. The bays are located immediately downstream from the intake and at the headrace tunnel inlet, while the forebay and spillway are at the end of the tunnel, penstock and powerhouse.

Sediment in the Pativilca River Basin

The Pativilca River Basin spans an area of 2,974 km2, with a mean elevation of 3,360 m above sea level (MASL) and river gradient of 2.1%. The river carries a large quantity of suspended solids, mainly due to the very active external geodynamics in the “medium” segment of the basin – that is, the area downstream from the river’s origin, where there is more of a gradient and strong external geodynamic activity. Sediment ranges from silt to boulders half a meter in diameter, which mainly originate in a mass erosion area 25 km upstream from the Cahua intake at Jelleragra Creek.

In 2013, sediment concentrations of up to 25 g/l were measured in the Pativilca’s intake area. The abrasion rate of the hydromechanical equipment is very sensitive to the size of sand, defined here as particles ranging between 0.062 mm and 2.0 mm in size.

Cahua’s eight desander bays are are located immediately downstream from the intake and at the headrace tunnel inlet.

The abrasion rate is directly associated with the number of particles colliding against metal. It should be noted that the damage caused by each collision is related to the momentum of the particle, which is directly proportional to the mass (volume) of the particle, which is associated with diameter cubed, expressed as:

Equation 1

Vol. = 4/3π(D/2)3

Consequently, control of the maximum particle diameter is of the utmost importance. Failure to keep the total solid concentration rate steady may result in a dramatic change in the abrasion rate.

As a result, abrasion should not be seen as a gradual process but as a damage occurring in periods when the sand is bigger in size and there is more sediment.

Intake and desanders

The intake for the Cahua plant is located on the left bank of the river. The orientation of the intake and the pier on the opposite side convey the water flow directly toward the intake. This area of the river is characterized by high turbulence, most notably during floods when sediment concentrations are higher, which results in sand particles remaining in suspension and flowing into the desander.

The water conveyed by the intake tends to flow at a higher velocity along the right bank than the left bank, which results in a hydraulic imbalance in the channels that convey water to the desander bays. Each channel is designed to feed half of the facility’s eight desander bays. Each desander bay is 5.4 m wide and 50 m long and designed for a 24 m3 per second flow, or 3 m3 per bay.

There are extensive records on the suspended sediment concentration in different parts of the headworks, including the intake, intake channels and headrace tunnel inlet. A laboratory, solid gauging equipment and duly-calibrated sensors are used to conduct simultaneous measurements at the intake, intake channels, desanders, tunnel inlet and forebay. Granulometric and mineralogical analyses are also conducted as part of these annual measurements.

Simultaneous measurements were conducted for the first time in 2014, primarily to determine efficiency of the desander bays. This efficiency varies from 3% to 45%, with an average efficiency of 20%. The concentration of suspended solids in the forebay decreased by about 20% compared to the volume in the desander bays during the testing, which seemed to indicate that a portion of the solids not trapped by the desander build up in the headrace tunnel and bottom of the forebay.

Sediment caught in the desander was primarily comprised of sand, which accounted for 75% to 95% of material trapped, while the material carried into the forebay was primarily composed of silt (82% average).

These analyses also indicated the maximum diameter of the sediment carried into the forebay varied primarily from 0.84 mm to 2.0 mm.

Mineralogical analysis indicates that the material carried into the desander bays and forebay was mainly composed of quartz at a 78% concentration. Quartz has a low specific gravity, which explains its presence at the forebay. In addition, its hardness – 7.0 on the Mohs scale – makes it highly erosive to the turbine.

Other notable sediments included augite (12%), hornblende (7%) and mica (2%).

Turbine erosion

Many hydropower plants are affected by large amounts of sediments during the rainy season. The Andes in South America is an example of a region where there are hydro turbines that experience heavy sediment erosion every year. Some of the power plants owned by Statkraft in both Peru and India, Albania, Chile and Nepal have a large sediment load during the rainy season and the company is working closely with technology developers to find a sustainable solution that avoids the sediment erosion on the turbines.

Cahua’s original turbines were designed to operate with a solids concentration of 3 g/l. However, the suspended solid content carried by the river in the flood season largely exceeds this figure.

The original turbine runner is affected by both sediment erosion and cavitation. Cavitation can occur in heavily eroded areas, and this has caused the erosion to accelerate. In the original turbine, the cavitation occurred at the suction side of the stay vane at the runner’s inlet. The runner was also heavily eroded at the stay vane outlet and the outer diameter. The leaking water has also caused erosion in areas around the labyrinth sealing.

The sediment erosion has caused Statkraft Peru to have to annually refurbish both turbines, which costs about US$580,000 and 20 days of downtime per year. The refurbishment was carried out by Cahua plant personnel and included welding, grinding and machining each component. The power plant was equipped with extra turbine components, meaning the downtime of the plant was minimized.

One of Cahua’s original turbine runners shows significant wear caused by sediments in the Pativilca River.

As a result of this downtime, Statkraft decided to install new runners that can operate with sediment concentrations of up to 10 g/l. Manufacturers have developed a number of new materials and coatings, with the most successful solution seeming to be tungsten carbide-based coatings. In Francis turbines, the coatings have been applied to the cover, guide vanes and some parts of the turbine runner.

The first stage – upgrading the runners – began in March 2009 and involved replacing one runner with a unit designed and produced by Norwegian manufacturer DynaVec, while the other unit was left with the original runner so the new runner could be compared against it.

The upgraded unit generated 13.1 GWh-worth of extra production compared to the non-upgraded unit between March 13, 2009, and June 9, 2009, while the sediment during the period equaled 131,000 tons, with a maximum sediment concentration during operation of 20 g/l.

The upgraded runner, removed from service June 9 for inspection, did not have any visible erosion on the runner vane surface. The coating thickness had only been reduced by a maximum of 30%. This was a very promising result for the runner. However, the coating had some weak spots at the outer diameter where there is a sharp corner. Here, the coating was removed and some erosion was visible. The areas around the leading and trailing edge of the runner vanes at the hub and shroud are also weak points due to sharp corners in the slot where the runner vane fit in.

The second runner was subsequently replaced by DynaVec based using results from the first test. The main improvement included the application of Belzona 1321 – a ceramic-filled epoxy coating – to reinforce critical areas of wear observed in the first stage runner.

Intake and desander operation

In addition to the change of runners, several studies were conducted between 2012 and 2014 aimed at reducing the sediments reaching the hydromechanical equipment. The studies were primarily focused on the implementation of operating rules and alternatives put forward to increase the desander efficiency.

With regard to the intake, the inlet gates at the intake channels that connect to the desander were fully closed at all times, which prevented them from being used for flow control purposes – especially when relatively high flows occurred that gave rise to imbalances between the left and right banks of the system into the desander. Taking into account that an orifice is a better way to control the flow, it was decided to improve regulation into the desander by using the inlet gates to create orifices at the bottom of the intake channel instead of using the radial gates in the river.

Additionally, the desander bays had been operating with an overload, reaching values about 4 m3 per second instead of their designed 3 m3 per second. The first approach with respect to the desanders was to eliminate the hydraulic overload, which is was evidenced by the measurement campaign conducted in 2014. It should be noted that eliminating overload is the easiest and most effective way to immediately increase the operating efficiency of the desander without making changes to the structure.

Moreover, during events with heavy sediment loads, Statkraft Peru personnel made sure that the sand level at the desander bays was maintained at low levels. The desander bays are not deep at 2.5 m, and with sand building up it would have been possible for the desander to allow the transport of sand into the tunnel.

Additionally, different options were studied to improve the desander efficiency, including elongated spillways and permeable barriers – both of which are under study.

Thus far, improvements implemented include gate regulation to prevent turbulent flows in the desander, elimination of hydraulic overload in all the desander bays, and restricting inflows in each bay to the design flow of 3 m3 per second.

Conceptual modeling of rainfall-flows-suspended solids

In addition to the steps mentioned above, a supplementary and relevant aspect of controlling sediment damage to the powerhouse is the making of operating decisions at Cahua during the flood season. This can be facilitated with an early warning system for high solid concentrations at the intake. To this end, it is first important to have a model of rainfall-flows-suspended solids.

Prior to development of the model, it was necessary to consider the components and process of the erosion system, which is illustrated by the Preston and Schmidt model. For Cahua, only the short-term scale – that is, just a few hours – is of interest.

To define the guidelines of the conceptual model, it was necessary to consider the following questions:

  • How are system elements defined?
  • How are discretization elements connected?
  • How are relevant hydrological and sedimentological processes defined?
  • How is previous knowledge of the basin’s physical conditions integrated into the model?

The proposed model is composed of two sub-models: rainfall-flow, and flow-suspended solids.

To determine the first model, several direct relationship models will be analyzed given their simplicity, and because its mathematical formula prevents that at any time the rainfall is exceeded by the runoff. The models that Statkraft Peru proposes to analyze are Budyko, Countagne, Grusnky, Penuelas, Pizarro, Turc and Turc-Pike. The model to be adopted would be the one with the best performance in statistical tests.

Norwegian manufacturer DynaVex was responsible for providing replacement runners for Cahua – the second of which was further improved with the application of a Belzona ceramic epoxy for added protection.

The second model would be a probabilistic model with correlations between the flows and suspended solids measured at the intake. Once rainfall has occurred, the flow range that would occur at the intake would be determined using the first model, which would be the input variable to estimate suspended solids in g/l.

The tributaries in the Pativilica River Basin with the most generation of solids have been identified, and Jelleragra Creek has been classified as the representative creek in view of its location in the basin and its almost immediate reaction regarding solid generation caused by rainfall. In this area, a digital rainfall station will be installed with a remote signal to Cahua, providing an early warning of the presence of solids at the intake in order to facilitate operating decisions at the plant.


In view of the high content of solids in the Pativilca River, a number of activities have been undertaken, resulting in a reduction of aggregate unit downtime, increased working hours of turbines and a recovery of generation revenues, primarily due to the installation of new turbines and a new way of operating the intake, gates and desander bays. The payback period is expected to be about seven years, with the profits from recovered production from each runner equaling more than $101,300 per year.

Gauging and solid analyses were performed to know the properties of the sediments, and with these results, the following improvements were made:

  • A system that includes a small laboratory was installed for measuring suspended solids entering the desanders and headrace tunnel;
  • Turbines were installed that can work with 10 g/l of material in suspension; and
  • The operation of gates and sand traps was optimized.

The improvements achieved to date have resulted in a reduction in down time to seven to 15 days per year, longer working hours for the turbines and the recovery of revenue generation. While sediment control has seen a significant improvement, the solid management can still be optimized to increase revenues at Cahua through:

  • Arrangements in the desander bays, which are currently under review; and
  • An early warning system to operational decision making, using conceptual rainfall-flows-suspended solids modeling guidelines presented in this article.

In addition, tributaries in the Pativilca River Basin with higher solids were identified, and to further improve sediment management at the Cahua project, a digital rainfall station with remote signal is planned to be installed in September in these tributaries to give early warning (two to three hours in advance) of the presence of suspended solids in the intake for making operating decisions.

For this, a new model of rainfall-flow-suspended solids will be developed. Once the model is calibrated and according to the results obtained, the implementation of an early warning network will be planned. This system can then be replicated in other basins with behaviors similar to the Pativilca River.

Anibal Maita Espinoza is the deputy manager of hydropower planning for Statkraft Peru.

Related content

For more information about sediment-related issues, attend session 3J at HydroVision International 2016, where a paper will be presented entitled Holistic Sediment and Erosion Management.

More HRW Current Issue Articles
More HRW Archives Issue Articles

Previous articleHow Manufacturers are Approaching Technology
Next articleInnovative Techniques to Repair Remote Mountainside Penstock

No posts to display