it is nice to hear that someone is managing and making these type of decisions, and developing a strategy on how to best do this.  Many times, development around a reservoir for homes and recreation causes managers to operate the reservoir at a level that does not take advantage of lowering the reservoir somewhat in expectation of a severe flood.  With the increased possibility for weather forecasting, reservoir managers are not so apt to be surprised by a severe event.  But recreation interests and home owners do not appreciate periods of dam lowering for flood protection, and the lowering of water also affects potential for irrigation or other uses that are sometimes reliant on dam storage.  

To limit flood damage downstream however, details downstream need to be developed on the amount of damage expected as flood flows become more severe.  This involves mapping the extent of floodplain downstream and 100 year and 500 year floods are often used, but other flood estimates may be made if desired.  The damage analysis typically comes from knowing where damage is likely to occur.  It is unfortunate that many times, past urban and other development in many areas has been within floodplains, and in some instances perhaps, installing a reservoir for flood control may have encouraged this.  

It may be important in dam management to understand that the reservoir collects upstream sediment, and downstream has clean water effect which will tend to erode channel and banks for a substantial distance downstream.  So for some distance, the channel will have some increase in capacity, or at least not a decrease in capacity from too much sediment.  However, I suppose in extreme conditions, the downstream clean water effect could undermine steep, unstable banks that could lead to bank failure and added sediment.  It is going to be difficult to generalize what may happen in all circumstances.  

The main goal in dam management is to avoid conditions where the dam would fail, as this would likely produce damage even more severe that natural flooding.  Another goal would be to release a bankfull flow regularly, at least once per year or two years, to keep vegetation from encroaching on the downstream channel, which would potentially reduce channel capacity and encourage sedimentation by increasing flow roughness.  

Having a good communication system for flood management and excessive releases of flow should be able to help those downstream adjust somewhat, shore up levees, move mobile homes if located in floodplain, evacuate people if needed.  

Unless dams fail, they often reduce the impacts of floods, but they are not likely to eliminate flood damage.  In areas where extreme storms are forecast, releasing some prestorm storage downstream adds protection, so understanding how much to release could be estimated with some good hydrological modeling of expected inflows, reservoir storage at various water levels, and damage from various amounts of outflow releases. The damage analysis of various scenarios may also include the damage caused by lowering or maintaining the reservoir at a low level to help increase storage for flood protection, but reduce water availability in reservoir for other uses.  

I attached a large document on  small dams (60 MB).  It may have some further useful information and ideas.  There are various hydrological models that may help, and if you are not a hydrologist or civil engineer, you might seek out their assistance.  Weather forecasting along with various GIS data on geology, soils, land use, topography, climactic and storm history, flood mapping, etc. would be helpful in making reservoir management plans and possibly automating or programming tools to aid in decision making, public communications, emergency notifications, etc.

There may also be some ecological needs for dam releases also to consider that I did not mention.  Where releases can mimic natural events (hydrograph) instead of instantaneous increases and drops, the better for aquatic organism adjustment.  Also if people are fishing or boating downstream, abrupt changes are hazardous for them.  

Dear sir William F. Hansen

Thank you very much for your answer and your attached file.

It is really happy to receive your recommendation. You are very dedicated to this field.

In Vietnam, there are many large reservoirs built more than 20 years ago. During that time, in the downstream of the reservoir, the infrastructure has developed very strongly. The reservoir design criteria of 20 years ago are outdated and require new criteria for downstream safety.

The problem that we need to address is to find safe solutions for dams and downstream areas in extreme flood context. 

The methods we apply is hydrological, hydraulic simulation, and flood mapping in emergency scenarios.

From simulation results, we will propose reservoir operating procedures and other options to improve the effectiveness of reservoir flood mitigation and mitigation of downstream impacts.

However, with the 10, 20, 50 year floods we still need to find the criteria to optimize the benefits between flood discharges and downstream inundation.

Please share your experience in this matter. Thank you!

An internet search should bring more examples and expertise to consider.  I heard in Egypt, that sometimes they have built a dam within a dam, so if failure occurs, not all water is lost, and damage not quite so severe.  If your existing dams due not include bypass, relief channels to prevent overflow, then that might be a first step.  20years is not an old dam, except if designing for 10-50 year flood.  In the instance of multiple dams, failure of one may cause failure of all downstream, resulting in extreme damages.  You sounds like you are on right track, just make sure you try to validate the hydrologic models, and adjust for land use changes.

i agree with Hansen

I think that there are two areas which are of concern - the normal operation of the dam for downstream flood risk management and other purposes (hydropower, irrigation etc) and the safety concerns about the structure itself in extreme conditions.

Obviously the flood hydrology needs periodic review as more streamflow and meteorological data become available together with long-term projections in future climate.  The impacts of change are best assessed through hydrological and flood flow modelling as recommended above by others.

A good place to start for information is the International Committee on Large dams (ICOLD) see http://www.icold-cigb.net/. 

For current dam safety knowledge you could look at information from the US Army Corps of engineers dam safety programme.  Some years ago there was a European research initiative on this topic and the outcomes are still online see http://www.impact-project.net/

I hope these comments are of help.