CFD Modeling Provides Value Engineering

By Amir Alavi, Don Murray, Claude Chartrand and Derek McCoy

Value engineering follows a structured set of procedures designed to seek out the optimum value for long-term investment. Value engineering analyzes the requirements of a project for the purpose of achieving and optimizing the essential functions at the lowest total cost. This article outlines the optimization process followed in the development of a well-designed water intake facility with the aid of computational fluid dynamics (CFD).

The function of the intake is to divert the discharge from the headpond into the power tunnel intake under controlled conditions while minimizing the hydraulic losses, ensuring that no air is entrained into the flow for all conditions of discharge, and controlling the amount of sediment entrained in the flow to minimize wear on the turbine-generator equipment.

Air entrainment was evaluated using U.S. Army Corps of Engineers design criteria as provided in Guidelines for Design of Intakes for Hydroelectric Plants. In addition, the results of the CFD analysis were evaluated for signs of tendencies for air entrainment to occur, such as vortex generation.

The CFD analysis provided value by increasing the hydraulic performance of the facility and by reducing construction cost, enabling Innergex Renewable Energy Inc. to balance hydraulic performance risk with reduced construction cost.

Project description

The Lillooet River is in the Pacific Range of the Coast Mountains and flows into Harrison Lake. The proposed powerhouse discharges into the Upper Lillooet River about 60 km northwest of Pemberton, British Columbia.

The 81.4-MW Upper Lillooet Hydroelectric Facility comprises the following components:

– Diversion weir, spillway, sluiceway and intake situated about 500 meters upstream of Keyhole Falls on the Upper Lillooet River, at the head of an incised canyon;

– Water conveyance system consisting of an excavated tunnel about 2.5 km long with a 5.5 meter “D”-shaped cross-section, a downstream portal and a section of buried penstock, about 1.6 km long and 3.6 m in diameter;

– Instream flow requirement release facility to provide for a minimum instream flow release ranging from 2.3 cubic meters per second (m3/sec) to 4.5 m3/sec throughout the year;

– Powerhouse with four horizontal Francis turbines and switchyard situated on a natural bench above the Upper Lillooet River;

– Transmission line from the project area to the point of interconnection with the BC Hydro Grid just north of Rutherford Creek; and

– Permanent access roads to the intake, downstream tunnel portal and powerhouse areas.

The plant is expected to operate continuously up to and including the 1:10 year return period flood, at which time the generating units would be shut down to limit the amount of entrained sediment entering the conveyance system, including the turbine water passages. The design basis for the project is the 1:200 year return period flood, which is the limit of expected normal function and performance of the facility with no major damage.

The maximum credible basis for the project is the 1:500 year return period flood, where loss of equipment is possible and damage to structures may occur. However, under this scenario the structures are not to collapse and major components may be repaired or replaced.

The forebay control structure (FCS) is designed to divert the maximum plant flow to the power tunnel intake under controlled conditions and to divert the required instream flow to the downstream river reach under river discharges up to and including the 1:10 year return period flow. The FCS is designed to exclude debris (>150 mm) and ice from entering the immediate intake forebay and to minimize the entrainment of sediment. The FCS is designed to minimize hydraulic losses by providing a hydraulically smooth transition to direct and accelerate the flow from the upstream river to the power tunnel. The design requirements are based on geologic, hydraulic, structural and economic considerations.

The FCS includes an upstream inlet consisting of two bays, each 8 meters wide by 4.5 meters high, equipped with coarse trashracks. The invert of the coarse trashrack is set about 2 meters above the invert of the sluiceway approach channel so as to minimize the amount and size of sediment entering the immediate intake forebay area. The design and layout of the FCS has been assessed with the aid of CFD modeling (FLOW-3D), which progressed through a series of configurative refinements.

During passage of the maximum plant flow, the velocity through the inlet will be about 0.8 m/sec. Downstream of the inlet, the structure will decrease in width and deepen to transition the flow toward two sets of inclined fine trashracks with a clear bar spacing of about 50 mm. These trashracks are designed to exclude debris that would be detrimental to operation of the turbine components from entering the power tunnel and turbine water passageways. Each set of trashracks will be 5.5 meters wide by 7.1 meters high. Under full powerhouse discharge, the velocity through the fine trashracks will be about 0.8 m/sec.

Immediately downstream of the trashracks, a trough collects and flushes some of the finer sediments that pass through the trashracks. This collection trough is provided with a slotted plate and a gated pipe outlet that serves as the instream flow release facility. Downstream of the collection trough, the FCS transitions to the power intake structure to direct the plant flow to the excavated power tunnel.

The function of the power tunnel intake is to direct the plant flow into the power tunnel under controlled conditions while minimizing the entrance losses and ensuring that no air is entrained for all conditions of plant flow. A 5.5-meter-wide by 5.5-meter-high service gate and an upstream set of stoplogs are installed at the power intake.

The service gate is designed to close against the full plant flow in the event of an emergency and provides for inspection and maintenance of the power tunnel. Under normal operating conditions, the plant flow is controlled by the turbine wicket gates. The stoplogs enable dewatering of the power intake opening for purposes of inspection and maintenance. An air vent, located immediately downstream of the service gate, facilitates free air flow during emptying and filling of the power tunnel.

A sluiceway is located immediately adjacent to the inlet of the FCS. The sluiceway channel is 9 m wide at the upstream end of the FCS, reducing to 6 m wide immediately upstream of the sluice gate. The sluiceway channel invert slopes from elevation 657 meters at the upstream end to elevation 656 meters at the sluice gate location.

The function of the sluiceway is to discharge accumulated sand and silt deposits from in front of the FCS during normal operation. During flood periods, bed load may deposit at the entrance to the FCS. By operating the sluiceway at full capacity during floods, a current is maintained and the channel will be kept clear. During normal operation when most of the flow is diverted, a sediment bar may build up progressively in front of the FCS and eventually sediment could pass into the intake. This can be prevented if the sluiceway is opened periodically to flush the accumulated deposits.

CFD modeling software

A three-dimensional numerical model was used to assess and optimize the hydraulic performance of the headwork facilities. The FLOW-3D program, distributed and supported by Flow Science Inc., was used to carry out the three-dimensional modeling.

This program simulates the dynamic behavior of fluids in three dimensions through a solution of the complete Navier-Stokes equations of fluid dynamics. The program is capable of simulating free surface flows, including transitions between supercritical and subcritical flow, with a single model setup.

Before its application in this study, the model was carefully reviewed and tested to verify the performance of the model under a number of similar real world applications. In previous project designs where both CFD and physical models were undertaken, the results of the FLOW-3D model compared well with results of physical model testing that had been conducted, thus providing confidence in the model capabilities. For this reason, the FLOW-3D model can be used with confidence to confirm the hydraulic performance of the intake headworks.

The Upper Lillooet facility was modelled with the aid of CFD modelling only; no physically modelling was undertaken. In previous project designs where both CFD and physical modelling were undertaken, discharges agreed to within 3-5% of one another and flow patterns were also in general agreement.

FLOW-3D model set-up

A model of the headworks area was set up from 50 meters upstream of the FCS to 300 meters downstream of the sluiceway structure. The model represented the headpond area, headwork facilities and downstream natural river channel.

The spillway (using an Obermeyer gate) and sluice gates were modeled in the fully closed position. The total discharge was set at 57.5 m3/sec; 53 m3/sec for the plant operation and 4.5 m3/sec for the instream flow requirement. The FCS diverts the plant discharge to the power tunnel inlet portal. To be assured that the model captures the characteristics of the power tunnel hydraulics, the first 50 meters of the power tunnel was also included in the model set-up.

Care was taken in selecting the upstream and downstream boundary conditions of the model to ensure that the entrance conditions, approach losses and tailrace water levels would be correctly simulated.

The physical representation of the geometry was constructed using AutoCAD and then imported into the numerical model as a stereo lithographic (STL) file. The upstream boundary for the model was set at a constant water level of 664.1 meters, while the downstream boundary was set as a water level downstream of the natural river control.

The K- turbulence model used in the FLOW 3D program is one of several options, but has been suggested by Flow Science as it is the most widely used for real-world problems. It is a two equation model that solves transport equations for turbulent kinetic energy and turbulent dissipation rate. Near rigid boundaries in the domain (walls, topography, etc.) a no-slip condition is assumed where velocity at the boundary is zero.

The model was run for 500 seconds (prototype) to simulate the selected scenarios. This time frame was sufficient for the model to achieve convergence.

The downstream boundary condition is set to a specific water height outside of the influence of any potential backwatering effects – this is appropriate for all river modelling if the outflow is set far enough downstream as it has been done in this modelling. The upstream boundary condition is set as a mass flow through the upstream face of the model domain, which allows for changes in water elevation and flow patterns.

Convergence in Flow3D is evaluated using two methods. First, the total water balance must agree between model inflow and model outflow, which can also be similarly checked by looking at the total volume of fluid in the model. Second, the total kinetic energy of the modelled system must sufficiently stabilize.

Physical modeling was not completed for this study as previous project experience by Hatch validated the performance and independent use of the Flow3D model.

Scenarios investigated

The objectives of the CFD analysis were to maximize the hydraulic efficiency of the headworks by minimizing the hydraulic losses, to minimize the cost of the capital construction, to improve the constructability of the headworks, and to simplify the operation and control of the headworks as much as practicable.

A number of layouts were considered, starting with the original layout established during the feasibility study for the project. The upstream bay (see Figure 1 on page 28) is only contributing about 15% of the diverted discharge, while the center bay is contributing about 50% and the downstream bay about 35%. The arrangement and orientation of this layout is, therefore, ineffective, resulting in underutilization of the upstream bay as well as higher hydraulic losses due to the uneven distribution of flow, particularly along the left side of the FCS.

The CFD model of the initial arrangement and orientation was ineffective, resulting in underutilization of the upstream bay.
The CFD model of the initial arrangement and orientation was ineffective, resulting in underutilization of the upstream bay.

The next scenario consisted of an intake structure with only two bays, eliminating the upstream bay. This arrangement resulted in a higher flow velocity through the structure while still showing a partial dead zone along the left side, although not as large of an area as compared to the previous layout. The flow distribution was improved to about 40% to 60% through each bay, but still not ideal. For the most part, the velocities through the upstream inlet as well as through the intake are higher than desirable, resulting in excess hydraulic losses.

To eliminate the apparent dead zone along the left side the inlet structure, it was decided to skew the alignment of the FCS by some 60 degrees. This resulted in a significant improvement to the flow distribution between the two bays.

The average flow velocity was reduced as better utilization of each bay was achieved with this arrangement. However, an apparent dead zone still appeared, this time along the right side of the inlet. With this arrangement, the flow distribution entering the power tunnel intake was satisfactory.

Now that the orientation of the intake structure was satisfactory, the study continued on improving the velocity distribution and magnitude through the FCS. First the inlet was widened from 6 meters to 7.2 meters to increase the area of the inlet to reduce the velocity. This reduced the velocity by 20% and still retained a relatively good flow distribution through the FCS.

With the next arrangement, the invert of the trashrack was lowered to further reduce the velocity. It was decided to increase the inlet area gradually to study the impact of widening and lowering the trashrack in increments. The lowered trashrack invert greatly improved the flow velocity and distribution through the inlet. The velocity was still slightly higher than desired but close to optimum.

A further increase in the inlet width to 8 meters improved the hydraulic flow conditions through the inlet, satisfying the velocity criteria through the FCS.

Additional streamlining at the upstream inlet improved the flow distribution at the inlet. The flow is streamlined throughout the FCS as well as through the inlet. The flow pattern and velocities at the entrance and in the approach channel are acceptable, with the reduced flow velocity at the inlet.

The reduction in velocity at the inlet will also result in a reduction of sediment entrainment to the intake, thereby reducing the ingestion of sediment into the power tunnel and downstream works. The performance of the sluiceway channel is satisfactory in its ability to flush accumulated sediments from in front of the intake. The hydraulic performance of the inlet and through the structure is satisfactory, with hydraulic losses estimated to be about 0.1 meters at full power plant flow.

Shear stress was computed in a separate CFD model of the sluiceway for the purpose of clearing sediment from the headpond. For the purpose of this intake CFD model, shear stress computation was not required.

By way of reducing the approach velocity at the intake, the potential for sediment entrainment is correspondingly reduced. The cell size varies throughout the model but around key features such as walls and transitions the cell size was 0.125 m in all directions.

It is presumed that this refers to the CFD mesh used. The mesh size used was to a sufficiently small size so as to balance computational intensity (model run-time) versus an acceptable tolerance of model error. As mentioned above, grid size was as small as 0.125 m in the vicinity of key features and transitions.


The ultimate layout of the headworks facility achieved the objectives of the CFD analysis. The hydraulic performance of the headworks was greatly enhanced, not only by reducing the hydraulic losses but also by ensuring satisfactory hydraulic operation of the headworks throughout the life of the facility.

The cost of the capital construction works was significantly reduced. The overall footprint of the headworks (see Figure 2 on page 30) was reduced, with less excavation and 35% less concrete required.

The CFD model of the final layout includes additional streamlining at the upstream inlet, improving flow distribution and reducing sediment.
The CFD model of the final layout includes additional streamlining at the upstream inlet, improving flow distribution and reducing sediment.

Overall, the operation of the headworks has been simplified and the CFD analysis provided confirmation on the satisfactory performance of the intake including its individual components as well as the operation of the adjacent sluiceway.

Value was achieved in at least three ways: reduced construction quantities, as mentioned above; eliminating the need for physical modelling by use of CFD; and accelerating the schedule by revising the preliminary design schedule.

Amir Alavi, formerly with Hatch Ltd., is now senior hydrotechnical engineer for MWH Global. Don Murray is senior hydrotechnical engineer for Hatch Ltd. Claude Chartrand is vice president of Engineering and Derek McCoy is engineering manager for Innergex Renewable Energy Inc.

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.

More HR Current Issue Articles
More HR Archives Issue Articles
Previous articleHydro Site Evaluation Project Could Aid Development
Next articlePerspectives: Problem Solving

No posts to display