This move to operational applications is a natural progression in the increasing use of modelling in water distribution. Understanding the drivers for this change requires a brief view of history as well as current imperatives. The key issues behind the trend are:
- The historical
- Hydraulic modelling
as a strategic planning tool
- Less water, more control
It will be shown that the importance of hydraulic modelling in operational management is likely to increase over the coming years as the water utilities are required to forecast and manage network performance with every greater degrees of accuracy in order to meet regulatory and customer expectations.
Mathematical modelling has been around for many years. It emerged as a formal science in the early 1940s, the period that also saw the birth of the computer. That may or may not have been a coincidence, but their relationship between the two has been close ever since, with the progress of the computer underpinning the changes in modelling.
Developments in computing have of course been staggering over that period, but can be simplified to two trends. The first is the incredible increase in computer power. For example, over 40 years the number of transistors on a chip has increased from 1000 transistors to 42 million.
Perhaps more interesting is the way in which that power has been put to use. Computer power has been directed to optimising ease of use and widening user access to the computer as much as to improving computing speeds. The advent of personal computers further broadened access and usability. And finally, simpler operating systems such as Windows opened access to a huge new constituency of PC users.
The development of modelling over this time has followed a similar path of increasing access and usability. The first network models were analogue models - wire networks and flows of electricity. In the early 70s digital models ran very slowly on mainframes, written from scratch by the modellers in scientific languages such as Fortran. Finally, the first full water modelling packages appeared, with concepts of connectivity, pressure, pipesize and simulation engine built in, and the use of modelling was opened up to a new constituency of engineers.
as an infrastructure planning tool
Most early modelling related to capital planning, probably for two reasons. First, that is where the most obvious savings are to be made. Optimising the design of a single multi-million pound infrastructure project may result in shaving a few percent off the total spend thereby giving a clear cost/benefit justification for modelling.
An equally important reason stems from the fact that strategic models do not have to be complex, and simpler models have the key attributes of being both quicker to run and quicker to build. The run time argument is obvious - there were times when even fairly simple models could take many hours to run, which challenged not only the patience of the modeller but the stability of the package and the computer. But the second argument of build time is even more important. Earlier modelling packages had few aids to model building efficiency, and because build, validation and calibration times were slower, only strategic models were cost-justifiable.
Early models were likely to be purpose-built by expert modellers for the specific capital project, run only by the people who built them, who would produce reports of results and then probably never use them again. The next project would relate to a different part of the network and need a different model. Re-use of models was rare.
So far the progress of modelling has been explained through what might be called the availability argument - it was there, and it worked. However, in recent years there has emerged an increasingly powerful requirement for using modelling.
Demand for water is now outstripping supply in many parts of both the developing and the developed world, leading to an increase in its value and political importance, and a renewed focus on the efficiency and productivity of water utilities. This, in turn, is partly responsible for the growth in regulatory control and accountability (also driven by factors such as privatisation and government legislation).
Further impetus has come from the new security environment the world now faces. Initiatives such as the US Homeland Security programme place further demands on water operations, requiring the evaluation of options and the establishment of the most effective and robust operational policies.
Factors such as these demand that water utilities understand the operational behaviour of their networks with utmost precision: to undertake modification works as cost effectively as possible; to minimise leakages; to respond to customer complaints in an informed and balanced way; and to anticipate disaster scenarios and plan responses.
as an operational decision-support tool
Since the day when modelling was used solely for simplified strategic evaluations, advances in modelling packages have enabled a very different sort of model to be built. The key characteristics of these models are:
- A high level of detail - very detailed models can be built because model build times and run times are constantly reducing. The value of detailed models, including all pipes, all valves, and all pumps for an area - is that they can be used for operational decision making. The old strategic models had so many simplifications and assumptions that, useful as they were for high level optimisation, they were not accurate at the operational level.
- Easy links to
all data sources - models can be easily linked to GIS data, customer files,
and other company and external data sources, ensuring good input data and
ease of input.
- Easy links to
output files - output to Word and Excel ensures that the results can be viewed
and understood easily, extending the circle of use way beyond the model-builder
to engineers with no specialist model knowledge.
- Good housekeeping
- in the past modelling was not conducted to the standards of other IT projects,
to their detriment. Good housekeeping of such issues as version control and
audit trails broadens the user base of models.
- Powerful simulation
engines that permit Extended Period Simulation - simple models running for
short time periods and examining only peak conditions are of limited value.
However, for a model to run accurately enough to support decision-making by
operations management, the modelling package must have a simulation engine
that runs big models quickly and reliably over extended periods way beyond
the 24 hour effective limit of smaller packages.
- Re-use of models - models are no longer discarded automatically. Many companies keep them continuously updated, ready for application whenever a relevant issue requires it to be run.
The fact that models can be re-used has the benefit that they can be proven to a depth that can never be achieved with a throw-away model. Once a model has been used a few times for different purposes and been found solid, it will be trusted by operational staff to support their decision-making. We list here just few of the possible applications that operational staff may address through models:
- Simulating Pollution
Water companies can use modelling to track the fate of any decaying or non-decaying substances, which might have been introduced either accidentally or intentionally, as they move through the distribution system. This understanding enables managers to assess the action that must be taken, identify the required response time and address the most critical areas first.
The model can then be used to simulate mains flushing to facilitate the effective removal of the pollutant from the system, minimising the disruption to normal supplies.
- Source Blending
By tracing the percentage of water received at any point in the system from any of the sources to the system, water engineers can track the hydraulic boundary between the sources, assessing the areas where source blending occurs. This can lead to identifying required system modifications, often based on model results and spatial data such as customer complaints for taste and odour.
- Pumping Systems
Pumping costs are one of the most significant investments made annually by a water provider, yet very few assess their system operation, thereby failing to identify potential savings on OPEX.
Models can be used to optimise pump operation, maximising pumping during periods of cheaper electricity, and enabling operations staff to assess the efficiency of the existing pumps and any need for replacement/refurbishment and the potential reduction in running costs that might result.
A well-calibrated model with accurate demand allocation can be used to assess system chlorination requirements. The water provider can assess the concentration necessary at the water treatment works / source and determine the requirement for any additional chlorine boosters within the system.
Accurate assessment of water quality including chlorine residuals can help the water provider reduce incidents of customer illness and customer complaints.
- System Storage
Models can be used to optimise the operation of storage facilities, balancing supply and demand to allow steady flow from a source, which in turn benefits treatment works¡¦ operators and bulk suppliers.
Optimising the use of reservoirs and tanks prevents excessive retention times, improving water quality and reducing incidents such as bacterial growth in the reservoir or in the distribution system. A calibrated model can be used to ensure adequate reservoir turnover, and to test system re-zoning if required.
- Pressure Management
A calibrated model can be used to assess the current system operating pressures and design pressure management schemes. This in turn can contribute to a reduction in system leakage, reduction in pressure related demand, reduction in burst incidents, and an extension in the life of the assets.
- Sediment Modelling
and Mains Flushing
Sedimentation Analysis enables the assessment of areas in the system that may be prone to sedimentation. This information can be analysed with mains material information and spatial data such as customer complaints of ¡§dirty water¡¨ to pinpoint the reasons for the system problems and to optimise the use of mains flushing gangs to alleviate the problem.
A model can also simulate hydrant flow to develop an effective mains flushing programme, and use property address data held with the network to list customers affected by the flushing programme.
The argument above boils down to a simple summary. First, the need for efficient and effective operational management of water distribution has never been greater, and this need is likely to grow in the future. Second, modelling packages have never been better placed to contribute to meeting those needs. The final ingredients required is the human dimension. Operational engineers need to:
- understand the
opportunities that modelling offers them for solving the issues they have
- trust that the models they have access to - pre-existing models that can probably be applied to their projects with little change - represent the engineering reality of network operation, and not a mathematical abstract.
The first of these is a matter of the momentum of the industry - the issue is being discussed more, the body of successful case studies grows, and the potential is clear to most operational engineers. However, that still leaves the issue of trust - hearts rather than minds. To progress towards hydraulic modelling in operational management, operations staff need to talk with the experienced users of their company¡¦s models, understand the extent to which they are proven and reliable, and if the answers are positive, press ahead and reap the benefits.
This paper was first presented by Paul Banfield at the Water and Wastewater Europe Conference held in Barcelona, Spain, 25-27 May 2004