data analytics, Data to Insight, Forecasting

Demand Forecasting and Rate Sensitivity: An Introduction

WaterFront is proud to bring you a new series on Demand Forecasting and Rate Sensitivity. This first post is a light introduction to the topic and will follow with more technical examples via case studies from John Cook and Edwin Roehl of Advanced Data Mining (ADMI). I hope you enjoy.

Why Demand Varies

Customer demand influences virtually every aspect of the utility’s operations. It controls daily operational decisions such as storage tank fill rates and peak hourly pumping rates, the daily quantity of water supply treated and the quantity of chemical usage.  It further influences capital needs such as, How large must water mains be sized?  Storage tanks?  Pumping stations?

It drives the question, When will it be necessary for the utility to expand its water or wastewater treatment plant or have to borrow money to pay for capital improvements?  It influences decisions regarding such long-range strategic decisions such as source of supply.  When will a new reservoir or water intake need to be constructed?   What is the reservoir safe yield and how long will its supply be sufficient to meet customer demand?

What influences customer demand?  It is influenced and controlled by customer choice, e.g., “Should I wash the car today?”, and varies with time-of-day, day of the week, the season, the weather, water and wastewater rate structures, and changes in the customer base. Variable demand causes variable production and revenue, which together make management planning and decision making more difficult and risky.  A utility’s service area may experience sustained rapid growth. Or, a region being served may experience unprecedented drought or excessive rainfall. Meteorological forcing is known to be an important cause of variability for most service areas. Consequently, the utility may make significant over-investments in infrastructure and expanded production capacity without having the customer revenues to adequately support the increased investment.  This problem can also influence the cost of future capital with a hindered ability to pay debt retirement and a lower overall credit rating on future debt instruments.

Benefits of Forecasting

The benefits to demand forecasting should be obvious based upon the above considerations. The utility will wish to make important adjustments to wholesale and retail rates to accommodate changes in demand to stabilize its revenue stream.  In addition, a quantification of the sensitivity of demand to both weather and rate variability is most helpful. Sensitivity analysis quantifies the relationships between a dependant variable of interest and causal variables, e.g., we know demand is somehow dependant on ambient temperature and precipitation. Computing sensitivities requires defining the relationships between variables through modeling. Models generally fall into one of two categories, deterministic and empirical. Deterministic models are created from first-principles equations, while empirical modeling adapts generalized mathematical functions to fit a line or surface through data from two or more variables. Calibrating either type of model attempts to optimally synthesize a line or surface through the observed data. Calibrating models is made difficult when data has substantial measurement error or is incomplete, and the variables for which data is available may only be able to provide a partial explanation of the causes of variability. The principal advantages that empirical models have over deterministic models are they can be developed much faster and are more accurate when the modeled systems are well characterized by data. However, empirical models are prone to problems when poorly applied. Overfitting and multicollinearity caused by correlated input variables can lead to invalid mappings between input and output variables.

Once the sensitivity to rate changes and schedules are understood, rate schedules can be designed to help maintain a stable revenue structure as well as a steady water demand.  A steady water demand is one in which peak hourly rates are more commensurate with peak daily flow rates and seasonal flow rates.  Having a stable water demand enables water plant and distribution system infrastructure to be expanded in a more precise manner—not too soon and not too late.  This makes for an optimal use of capital funding and operational resources and helps to ensure that infrastructure is properly sized.  It enables planning to be conducted with a degree of stability that heretofore would not be possible, and enables the utility to literally weather the storm of a reduced revenue stream.

Advertisements

About noahmorgenstern

Entrepreneurial Warlock, mCouponing evangelist, NFC Rabbi, Innovation and Business Intelligence Imam, Secular World Shaker, and General All Around Good Guy

Discussion

2 thoughts on “Demand Forecasting and Rate Sensitivity: An Introduction

  1. I like it.

    Posted by jaimesal | April 19, 2012, 1:38 pm
  2. My experience is that rates are a very weak influence on water and sewage ‘demand’ /volumes – both are still very cheap in general terms, and external factors such as rainfall are much greater. In the market I’m familiar with, several utilities were required to produce rolling 12 month demand forecasts each month. Very different methodologies were used – including complex (and expensive multi-variable stochastic models). But the utility that forecast the same number every month, with a 10% bump for the summer holiday, was just as (in)accurate as any of the other models. Rainfall was far and away the biggest influence, but with about a two-month lag. This is an environment where consumption rates are already very low, however, after several years of drought, so the base level of demand is approaching a non-discretionary level.

    One of the underlying problems is that an industry like water, with a highly fixed cost structure, is continuing to expose itself to substantial revenue risk as a result of the belief that rates are an effective demand management tool. The evidence base doesn’t support that in many environments.

    Posted by Richard | February 3, 2013, 12:17 am

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: