PREVENTING FRAZIL ICE ACCUMULATIONS AT HYDROELECTRIC FACILITIES

Transcription

1 Ice in the Environment: Proceedings of the 16th IAHR International Symposium on Ice Dunedin, New Zealand, 2nd 6th December 2002 International Association of Hydraulic Engineering and Research PREVENTING FRAZIL ICE ACCUMULATIONS AT HYDROELECTRIC FACILITIES Brian Morse 1 and Tung-Thanh Quach 2 ABSTRACT For a 3.5-km open reach of river upstream of a hydro dam, we estimate daily heat loss to the atmosphere during typical winter conditions to be 4000 MWh, or 166 MW (thermal). This heat loss results in frasil ice formation equivalent to cubic meters of ice per day. This ice continues downstream and flows under a fixed cover upstream of the dam causing a hanging dam and potential flooding. In the past, fortunately, groundwater and wastewater coming from multiple sewers emptying into the river offset the heat loss. However, in a concerted effort to clean up the aquatic environment, the local government recently diverted these contaminated waters to treatment facilities downstream. In the case of Rivière-des-Prairies, skirting the north shore of Montreal, Canada, the diversion of 3.3 m 3 /s of water having an average winter temperature of 9 ºC resulted in loss of at least 124 MW. So after the treatment facility was built in 1995, frasil ice production in the reach quadrupled due to heat loss from 42 MW to 166 MW. To prevent upstream flooding, the hydroelectric facility is forced to perform aggressive ice interventions under very stressful circumstances. It also results in a loss of electrical power (due to the necessary flow diversion and the loss in head due to ice accumulation in the reach downstream of the dam). In this paper, we discuss the effectiveness of various intervention strategies, including: (a) using heat exchangers to recover some of the heat from the waste water, (b) using local ground water in a closed loop to heat the river, (c) pumping treated water from its downstream location back upstream of the dam, (d) installing traditional ice booms and installing tension ice booms (that have a net floating on the river surface on the downstream side). INTRODUCTION This presentation is complementary to that presented by Quach et al. at this conference. The Ottawa River (watershed of about 140,000 km 2 ) joins the St. Lawrence River at Montreal, Quebec. As it does so, it splits into three principal reaches. The centre reach is known as Rivière-des-Prairies that begins at the Ottawa River at km-44 and meets 1 Université Laval, Sainte-Foy, Québec, Canada, Brian.Morse@gci.ulaval.ca 2 Hydro Québec, Montréal, Québec, Canada, quach_thanh.tungydro.qc.ca

2 the St. Lawrence at km-0. Montreal, the largest city of Quebec is on its right bank and Laval, the third largest city is on its left bank. Except for a tiny stretch, from km-44 to km-28.8, Hydro-Quebec (HQ) forms a stable ice cover by installing 2 series of ice booms (one of which is at km-28.8). At mid-reach ( km-22 ) HQ operates a 50-MW run-of-the-river hydroelectric dam known as Centrale des Prairies. Immediately downstream of the dam is a 1-km long steep, rapid-flow reach that is considered to be an important fish habitat. Immediately upstream of the dam (from km-22 to km-25.3) there is a reservoir-type reach that easily forms a stable ice cover during the winter months (particularly since HQ installed some anchors to keep it in place). From km-25.3 to km-28.8 there is a 4-m deep, 270-m wide problematic reach that is the principal subject of this paper. The water in this reach flows too rapidly; it is difficult to form a stable ice cover. As such, it generates frazil ice during the winter months. This frazil ice accumulates under the ice cover in the reach immediately upstream of the dam. The accumulation increases head loss and forces HQ to take preventive measures to severely control water levels as expected, urban sprawl reaches the river banks on both sides so that there is virtually no freeboard left. Any error in dam operation would lead to an immediate flooding of houses, public institutions and municipal infrastructure. HQ is forced to operate a mini icebreaker that keeps a 60-m wide channel open from km-22 to km This channel acts as a sluice between the frazil reach and the dam s overflow spillway. Consequently, operational headaches abound, water is spilled unnecessarily and ice accumulates downstream reducing head. Should HQ ever loose the upper hand on the frazil ice evacuation process, flooding would be virtually immediately felt upstream. On top of it all, the sluice channel itself contributes additional frazil that aggravates the whole problem. The design flow of the facility is 800 m 3 /s. Rarely does Rivière-des-prairies receive less. The sluicing process requires the loss of about 80 m 3 /s and reduces the operating head from 8 m down to about 6.5 m. Since power produced is proportional to the flow rate Q and the head H, in all, frazil ice causes the loss of about 10 MW or about 25,000 Mwh over the 100-day winter period. Depending on the price of electricity, this represents a loss of 0.5 to 2.5 million dollars per year on top of which must be added the 0.3 million dollars a year of icebreaker operating expenses. Strangely enough, before 1995, the problems were much less severe. We believe that this is because at that time, the warm (8 10 ºC) water from the municipal sewers was diverted from the reaches upstream of the dam. During the 1980s and 1990s, the Quebec government made a priority of treating its municipal wastewater. In an effort to clean up Rivière-des-prairies, for the City of Montreal on the right bank, the Décarie- Rimbault collector (Q = 2.0 m 3 /s) is now diverted to a deep interceptor that leads to a water treatment plant at km-0 and for the City of Laval, it built the Marigot interceptor (Q = 1.3 m 3 /s) that leads to the La Pinière treatment plant at km Since the heat flow rate Q e is

3 Qe = Ctσ QTe (1) (where C t is the specific heat of water = 4186 J kg -1 ºC -1, ρ is the water density = 1000 kg/m 3 ), Q is the flow rate = 3.3 m 3 /s) and T e is the temperature of the wastewater = 9 C)), the sum of these discharges correspond to a diversion of 124 MW from the river upstream of the dam. We note that whereas the Quebec government requires strict environmental assessments prior to intervening in surface waters, in this case, in their desire to improve the river water quality, the impact of ice on the flow regime was sadly overlooked. The loss of heat Q a from the water in the open reach (km-25.3 to km-28.8) to the atmosphere can be estimated by the following equation: Qa = Ca ATa (2) where C a is a coefficient of energy exchange between the flowing water and the atmosphere. This value which is in function of the wind speed, the relative humidity, hours of daylight, the water current speed etc. has been evaluated by Marcotte (1975) for the Montreal region as 24 to 31 W/ m 2 depending on the month, A is the open surface area of the reach (0,954 km 2 ) and T a is the average air temperature. Taking December as an example, between 1975 and 1984, the average air temperature measured at Dorval airport (some 10 km from the area of interest) is T a = 6,2 C. The average heat loss is therefore Q a = 148 MW. Knowing that there is much more heat loss on cold starless windy nights than on calm cloudy days, we can apply this gross analysis to the four principle winter months of Rivière-des-prairies: Table 1: Average heat imbalance before and after waste water diversion Average air temperature T a ( C) Net heat loss Q a Q e (MW) before waste water diversion Heat loss Q a (MW) since waste water diversion Increase in frazil production (%) since diversion Month December 6, January 11, February 6, Mach ,8 Nil This table demonstrates that prior to 1995, December produced very little frazil and by the time mid February came around, frazil ice production was all but shut down. Historically, the really problematic period was the end of December to the end of January. A more detailed presentation of this data is provided on a daily basis for many years in the figures presented below. So, Hydro-Quebec is stuck with the problem the bill and the threat of flooding the core metropolitan area of Quebec. The status quo is not acceptable. Our mandate is to confirm the suspected the source of the problem (loss of waste water heat) and to evaluate the feasibility of various options to remedy the problem.

4 REMEDY OPTION NO. 1: FINDING A NEW HEAT SOURCE If we could get some heat back into the system, the frazil problem could be attenuated. Our first potential heat source is ground water. For ecological reasons, we do not presume that our local government would let us pump ground water directly into the river but we do think that it may be possible to pump the water through a re-circulating pipe that would act as a big heat element to the river water as it flows pass. The concept is presented in Figure 1. Unfortunately, the geological investigation reveals that the hydrogeological yield (about 3 s -1 ) is only one thousandth of that required. Furthermore, even if we could pump 3 m 3 /s through the pipe, it would have to be 1000 m long to for the exit temperature to be low enough (2 C) ensure adequate heat exchange between the warm (12 C) groundwater and the 0 C river water. Finally, the associated tunnelling costs are substantial. We therefore abandon this option. Laval Heat exchange pipe traversing river Rivière des Prairies Montréal Te = 12 C Underground conduit Aquifer : Te = 12 C at 100 m Te = 2 C m m Q = 3 m 3 /s 200 m Figure 1: Concept for heating the river using ground water The next option considered is to put a heat exchanger in the sewer interceptors. On the Laval side of the river, the interceptor passes quite close to the river at key locations. The idea is to excavate a water intake from the river to the interceptor, suck the heat out of the sewer water and release the subsequently warmed water back into the river. Assuming the exchanger provided 33 % efficiency, the amount of heat extracted would only be 16 MW. Considering the capital costs, the difficult of operating the water intake (due to frazil ice congestion) and the anticipated concerns related to contaminating the river water should there be leaks in the water exchanger, this option does not seem the most promising. A similar option could be used on the Montreal side. The collector is large (2 m) and it is possible to consider a 30 cm pipe leading from the river up the collector to the point where the water now falls into the Montreal interceptor (very deep at this point). This could potential provide 25 MW should an exchanger be installed in the Décarie- Rimbault collector before the water is diverted to the interceptor. The problem with this scheme is that this sewer water is already used as a heat source melting the snow removed from the streets. Therefore, given the cost, the hassles and the lack of temperature reliability, this source does not seem too promising.

5 We were therefore forced to examine the possibility of using the purified water leaving the Laval water treatment plant La Pinière. Its flow rate (normally at 8.5 C) varies between 2.3 and 2.8 m 3 /s. It could therefore theoretically provide 89 MW to our reach. To evaluate the impact of this alternative on frazil ice accumulations, we studied the daily air temperatures from 1975 to We assume the initial water temperature of Rivière-des-Prairies to be 0 C and therefore all the net heat loss is converted to frazil ice. We then assume that the ice would be evenly spread under the A d = 2.5 km 2 ice sheet immediately upstream of the dam. Figure 2 presents the ice accumulation based on daily temperatures averaged over the years 1975 to The conversion from heat loss (Q a -Q e ) to a daily equivalent increase in frazil accumulation Df (m) is made in the following steps: D f = (Q a -Q e )*S/(K i ρ i Ad) (3) Where S is the number of seconds in a day (86400 s/d), K i is the heat of fusion (333,000 J/kg), ρ i is the density of ice (910 kg/m 3 ). For example, for a typical December day, Q a = 148 MW. Under status quo conditions, Q e = 0 but the sewers used to provide Q e = 124MW and if we use heat from the treatment plant Q e = 89 MW. So the ice accumulation on a typical December day prior to 1995 would be D f = (148,000, ,000,000) / (333, ,500,000) = m. Over the month of December, the accumulation would be = m. 2,5 Frazil accumulation under ice cover 2,0 1,5 1,0 0,5 Status quo conditions with heat diverted from treatment plant with heat from sewers (prior to 1995) 0,0 2-déc 1-janv 31-janv 1-mars 31-mars Figure 2: Frazil ice accumulation upstream of dam Clearly, rerouting the outflow from the treatment plant does not restore ice accumulation to pre 1995 levels but it does go a long way to clearing up the problem. Normally ice accumulation would only begin in the second half of December and would be on the mend in late February. For the average year, the maximum thickness would be reduced by 56 % from 2.3 m to 1.3 m. We performed a similar analysis on a year-by-year basis. The table opposite shows that the efficiency in the reduction of the frazil accumulation can vary between 31 % and 79 % and depends on the severity of the year. Unfortunately, it is during % % % % % % % % % %

6 the coldest years that the improvement is the least efficient. In our opinion, this seems like a good solution. However, there are feasibility issues that need to be addressed. First of all, we expect that many would oppose such a plan because it is bringing sewer water (even though it is treated) back into their beautiful back yards. Secondly, there would be an environmental impact during the construction period of installing a pipeline through the municipal streets. One could always tunnel a route but the cost is about double at 20 millions dollars to be added to the capital costs of the 500 HP pumps and the related infrastructure. Annual maintenance costs should be minimal but would require 700 Mw h annually to HQ to power the pumps. From a strictly cost/benefit point of view, the scenario is marginal. However, from our personal perspective, to the cost/benefit picture must be added the stress of the HQ employees trying to stay on top of difficult winter conditions and the real threat of a flooding related lawsuit should they ever loose the upper hand. REMEDY OPTION NO. 2: PUTTING ON AN ICE COVER Of course, the traditional solution to frazil ice production problems is to promote the formation of a stable ice cover. Once a stable ice sheet is formed, it insulates the water from heat loss to the atmosphere and any ice aggregation stays in place causing no hanging-dams downstream. This ice control method is already used at two locations on the river upstream of the problematic reach and our mandate was to check if it was at all possible to find a way to install one of the more efficient ice booms (Morse, 2001a and 2001b) in this reach. Well, with the help of André Bouchard of Environment Canada we measured velocity profiles at about 20 cross-sections. We also adjusted the output of a HQ HEC-RAS modelling effort to account for ice cover conditions, different flow rates and different operational water levels upstream of the dam. However, based on the design flow rate of 800 m 3 /s, as figure 3 testifies, no matter whether we used the 0.08 to 0.12 Froude criterion or the 70 cm/s water velocity criterion, there is just no way a traditional ice boom is going to work. In fact, we calculated that the water level would have to be raised a minimum of 1.8 m to even have a chance of working. But even for this scenario, there would be no guarantee. For our analysis, we presumed that an ice cover required 8 consecutive days of low flows. We then analysed 74 years of data (1922 to 1995) to establish the probabilities. We found that that whereas 4 years of 10 the flow rate did remain below 800 m 3 /s for at least 8 consecutive days, there were also 4 years out of 10 where the minimum 8-day flow was as much as 1000 m 3 /s. So, even if it were possible to raise the water level 1.8 m (which would in fact cause urban flooding), the installed booms would only succeed in forming a cover every other year. We then analysed two more options: Reduce the flow rates from the Ottawa River (known as Lac-des-deux-montagnes locally) to Rivière-des-Prairies during the key ice formation period. This flow reduction could be achieved by installing a mobile dam at the Lake outlet. The most environmental friendly dam would probably be an inflatable rubber weir although Abdelnour (2002) suggests that it may be done using a head-loss boom. We note that HQ uses flow reduction very successfully a few km upstream from Rivière-des-Prairies on the St. Lawrence River at Beauharnois. In that application, flow is reduced at the upstream dam (Moses-Saunders) and the downstream dam (Beauharnois) to achieve a Froude number of What is less well known is that to

7 form the cover more quickly, a very cold period is targeted and at that time, lake ice already formed upstream is cut out using an ice-breaker and floated down into position immediately upstream of the Beauharnois intake. So this may be an interesting solution particularly since reducing the flow in Rivièredes-Prairies would help maintain a more acceptable flow in the neighbouring Rivièredes-milles-îles. However, one may be required to operate the dam more often than just during the ice formation period because of the relatively high frequency of high flows (1000 m 3 /s and more) in des-prairies. Unless one reduced the flows during these periods, one would run the risk of having the cover detach from its banks and piling up in a mini jam downstream. This occurrence could be potentially worse than the status quo. 0,24 1,10 0,21 1,00 No. de Froude (-) 0,18 0,15 0,12 0,09 0,06 0,90 0,80 0,70 0,60 0,50 Mean water velocity (m/s) 0,03 Estimated average velocity (m/s) Velocity criterion for ice boom deployment = 0.7 m/s Bridge locations Estimated Froude number (-) Froude criterion for ice boom deployment = , , , , ,5 25 kilometre along problematic reach of Rivière-des-prairies Figure 3: Velocity and Froude evaluation for traditional ice boom deployment 0,30 A final option for putting a cover on the problematic reach was to use a tension ice boom (figure 4). Since velocities are too strong to prevent floating ice floes from submerging underneath a cover, the idea of the tension boom would be to promote the ice to attach itself to a floating net that would be supported in tension by cables anchored to the river bottom and river banks. The principle of this kind of boom is that the net would provide a matrix to support ice sheet formation that would then freeze into the banks. Unfortunately, from a design point of view, one must dimension the cables to support the entire load. This load is considerable since there is no arching of the ice forces into the banks (as there is for a traditional boom). Secondly, whereas a traditional boom only needs to initiate ice cover progression, a tension boom only forms a cover where there is a net. This makes it extremely expensive to cover each square metre of river surface. So although we did not recommend this alternative, knowing

8 that the investment in the boom would not be lost (because it could eventually be used elsewhere) HQ found it sufficiently interesting to try it out this last winter. Abdelnour and Quach will certainly publish the results of the trial. CONCLUSIONS 1. The invasion of urban sprawl right up to the water s edge once again puts enormous pressure on water body managers. 2. In northern climates, environmental impact assessments must start to examine the impact of a project on ice processes. 3. There is a lot of thermal energy in sewer water during wintertime if we can find a way to exploit it. 4. In the case of a dam in the middle of an urban area, in addition to cost/benefit ratios, the financial and image threat of the hydroelectric utility would be too important to be ignored. 5. In this specific case study, there are two possible long-term solutions: divert the effluent of the treated sewer water from downstream or erect a mobile dam upstream. Of the two, the first seems preferable to the second because the result is more certain and because the local population would probably abhor the construction of yet another dam Distance downstream of steel cables (m) River flow Polyethelene nets supported by steel cables Distance from river left bank (m) Figure 4: Concept for tension ice boom REFERENCES Abdelnour, R. Head loss Generator for Water Control in Rivers and Lakes. Pamphlet by Fleet Technology Ltd. (2002). Quach, T. Thanh-, Tremblay, R-M. Abdelnour, R. and Morse, B. Assessment of Ice Control Procedures for the Operation of a Hydroelectric Power Plant. In Proc. 16th IAHR Conference, Dunedin, NZ (2002).

9 Morse, B. Theoretical Development of forces on cylindrical ice booms. Journal of Cold Regions Engineering 15(3): (2001). Morse, B. Dynamics of ice forces on booms. Cold Regions Science and Technology 33(1): (2002).