Demand Stream - Load Forecasting

Questions to be asked about future asset utilization include:

• How accurately has the organization predicted the demands for their service(s)?
• Has the organization produced an assessment of their future demands?
• Does this prediction take into account changes?
  • In unit demands
  • In the number of users.
• What allowances are made for weather predictions, if relevant?
• How has the organization performed in previous years? How close has the predicted demand performed against the actual demand experience?
• Does the prediction take into account segmentation in network systems, how small an area or category of asset is considered in the modelling process?
  • 5% of system
  • 10% of system
  • 20% of system
  • 25 000 population
  • 50 000 population
  • 100 000 population
  • 500 000 population.
• How are the customer usage records used in this process?
• Are customers dealt with in specific categories? for example Residential, Commercial, Industrial, Major Customers.
• Are strategic demographic changes and trends used in the modelling process?
  • Population per tenant
  • Dual occupancy
  • Offices being changed to apartments
  • Changing (for example: ageing) population in older suburbs.
• How is town planning and other economic development processes included in the demand predictions?
• Are future customer expectations taken into account in the demand predictions?
  • Quality of service
  • Water pressure
  • Voltage fluctuations
  • Number of interruptions and response times
  • Reduced noise
  • Environmental pollution
  • Response to complaints
  • Gas pressure.
• Does the organization model these demands into various scenarios with different probabilities? If so how many?
• How are these predictions linked to the actual assets and their current utilization?
• What is the design/demand criteria used to predict the date/time at which the current capacity will be exceeded? for example:
  • Peak hour
  • Peak day
  • Peak week
  • Peak month.
• How well is this process followed?
• How appropriate are these criteria? Do they relate to the current organization objectives
  • Regulatory environment
  • Industry standard
  • Economic viability.
• Does the demand and prediction include the use of any non-asset based strategies to manage the demand? for example:
  • Switch off (power programs)
  • Alternative sources/co-generation
  • Use water wisely
  • Peak day restrictions
  • User pays policies.
• Are there regulatory factors that are likely to impact on the organization and demand?
  • Greenhouse gas reduction costs
  • Fish ladders and variable level offtakes required on dams and weirs for environmental issues
  • Greenhouse gas reduction will push up cost of power and therefore reduce demand.
• What is the overall impact of changes in demand for services and how have these been addressed?
• What other strategies may impact on the demand for service? What will be the overall effect on demand?
• How accurately have these impacts been tracked?
  • Historical experience
  • Experience of others.
• At what level is the current demand or system utilization data accumulated? If demand utilization readings are obtained from lower levels in the system hierarchy or spatial distribution then the organization will have far greater understanding of the overall utilization and therefore the likelihood of capacity failures and the need to enhance or augment the system.
• How is the utilization monitored?
  • Meters that monitor pressure or other product demand quality attributes
  • Measures relating to customer expectations, eg service failures, their frequencies and impact on customers.
• Is the organization identifying all of the demand quality elements and are these matched to the customer's expectations? In many cases organizations have a bias towards capacity failures and fail to properly assess other customer expectations.
• What is the relevant quality and usefulness of the data collected from these monitoring systems?
  • Manual recording of data into hard copy system
  • Manual recording of data into electronic system
  • Automatic recording of data into electronic system by PLC into data logging system or full telemetry SCADA system.
• Is the data acquisition system capable of storing all the performance data for a period of:
  • 1 year
  • 2 years
  • 4 years
  • 6 years
• Is the data acquisition system capable of aggregating annual data into key categories and archiving this data for long term purposes
• Is the data acquisition system capable of producing average demand distributions, such as peak:
  • Day
  • Week
  • Month
  • Quarter
  • Average monthly distribution
  • Average annual distribution
• Is the data acquisition system capable of tying this distribution to other data that effects demand distribution, such as weather conditions, temperature etc.
• Can the system complete linear regression of this data to produce variations of the above outputs that show the demand against average or diverse weather patterns or other demand drivers?
• Is this utilization data reconciled against customer usage based meters? It is important to reconcile our system demands with what is really being consumed by customers. This allows the modelers to look at issues such as unaccounted product losses and other elements that may be useful in calibrating demand or predictive modeling.
• Does the demand forecasting model allow the group to model coincident demand probabilities? Can the organization model the differences between the various consumer or customer groups such as domestic, commercial, small industrial or large consumers? Does it have sufficient knowledge on the average and peak demand patterns for each of these major customer groups to be able to assess the probability of coincident peak demands impacting on the system?
• What is the quality of the modelling software in handling this demand information?
• Are the size of the assets loaded into the model or the number of nodes for the network being considered eg. nodal density
• How does the model accommodate demand information?
• Can the modelers show the true interconnectivity and redundancy of the system, allowing better calibration and more accurate predictions to be made?
• What are the Quality Assurance Processes followed to ensure the model is appropriately calibrated and truly reflects the performance of the system under known utilization factors? Are these QA Processes documented?