Zeroing In On Peak Demand

(Tip of the Month for January 2005, revised September 3, 2006)

Utility rates for peak electric demand are rising. Knowing when and how often peaks occur is the first step to cutting those costs. Central to that task is the creation of a load distribution curve using software such as Microsoft® Excel™.

What's A Load Distribution Curve?

Electric demand (in kW) varies over time. Most of us are familiar with load profiles that show such variation over a 24-hour period, with hourly (or quarter-hourly) demands seen in chronological order. When those hourly demands are instead sorted by their magnitude and charted by how often they occur (as a percent of total hours), a load distribution curve is created (see Figure 1).

What Does That Show Me?

Unless a facility has an unusually constant demand (e.g., an industrial plant with large and consistent process-related loads), its load distribution curve will likely look like the stretched out "S" seen in Figure 1. The upturn at the left side indicates a small percentage of hours with high demands while the downturn at the right side indicates a small percentage of hours at low demands. When Excel is used to produce such a chart, touching any point on that curve with a cursor reveals the load at that point and the percent of hours above that load. The tag seen in Figure 1 indicates that only 1% of all hours in the data sample are above 2261 kW, even though the peak load is 2470 kW.

Where high peak demand charges (e.g., above $10 per kW per month) are in effect, it may make economic sense to look for ways to trim those peaks. Determining the duration of such high loads helps quantify how much (and how often) peak shaving must be pursued to achieve consistent savings.

Here's An Example

A facility's peak load rarely exceeded 2261 kW: 1% of annual hours translates to about 88 hours a year. If demand is being monitored (by an automated system or an EMS operator), and a system is in place for responding to it (by either briefly curtailing nonessential loads or starting an on-site generator), controlling peak demand for only 88 hours a year should not be difficult.

Peak demand in most buildings is impacted by electric air conditioning during the 5- to 7-month cooling season. Highest demand charge rates (in $/kW per month) are often seen for the 4 summer months of June through September. As a result, the 88 hours in our example are likely distributed among those 4 months, and not throughout the entire year.

In some high-cost areas, a 200-kW demand reduction (from 2470 kW to 2261 kW) during the 4 to 6 months of the cooling season could save $10,000 to $30,000. It's important to remember, however, that demand charges may be based on the highest peak occurring during a month, so merely holding demand at 2261 kW once in a month may be insufficient to avoid such charges: demand must be held at or below 2261 kW during all of the daily peak periods in that month. An automated demand reduction system (or an alarm that alerts an EMS operator to take action) is therefore essential to ensure demand is controlled in real time.

How Do I Perform This Analysis?

To create the chart seen in Figure 1, start with a year of interval data from your electric meter or EMS.

Follow these steps:

1. To focus the analysis, eliminate data for months and hours when peak demand charges either do not apply or are low enough not to warrant attention (e.g., winter months, all hours between 6 PM and 8 AM); find those months and times in your utility's electric tariff.
    
2. Gather the remaining interval readings into a single column. Do not include data cells containing time or date information. If any data is in rows, use the "Transpose" capability in "Paste Special" to reorient such data into columns.
    
3. In the Excel toolbar, click on "Tools" and look for "Data Analysis…" in its drop-down list. If you don't see that option, find and open "Add Ins" in the "Tools" list and check off "Analysis ToolPak." If this option is already on your hard drive in a compressed format (typical for standalone PCs), it will be opened automatically. For systems connected to a network, a Windows CD or assistance from IT personnel may be needed to install this option. Once done, the Data Analysis tools will remain available for all Excel files on that PC.
    
4. Click "OK" and see "Data Analysis" appear under "Tools." Click on it to see a list of analysis tools.
    
5. Scroll down to "Rank and Percentile" and click on it. See its dialog box.
    
6. In "Input Range," enter the locations of the cells containing the single column of interval data (or just select those cells to make their location appear in the "Input Range" field).
    
7. Next to "Grouped By," be sure the radio button for "Columns" is chosen. Under "Output Options," choose "New Worksheet Ply" to create a separate worksheet containing the new chart (titling is optional). Click "OK."
    
8. Excel will create a four-column table with columns titled “Point” (a number you may ignore), “Column 1" (in which your interval data has been sorted into descending order, starting with the highest value), “ Rank” (another number you're unlikely to use), and “Percent” (showing the percent of values lower than the number seen in the same row under Column 1). Note: with many points to be ranked, Excel may take 20 seconds to 2 minutes to perform this analysis.
    
9. To create the Excel chart seen in Figure 1, populate an empty column with data that equal 1-[values seen in the Percent column]. Do so by entering "=1-F5" in the cell next to the first value seen in the Percent column (in this example, F5 is simply the name of the first cell in the Percent column). To replicate that formula in all cells in that column, use your cursor to grab and drag that cell's lower right corner down the column until you reach the last row in the Percent column. To the right of that new column, copy and paste the column of now sorted interval data seen under Column 1. Now select the data in the two new columns just created.
    
10. Open the Chart Wizard and choose the "XY (Scatter)" Standard Chart type.
     
11. Choose a subtype (the middle choice, "scatter with data points connected with smooth lines," is fine) and click "Next."
     
12. In Step 2 of the Chart Wizard, be sure the radio button next to "Columns" has been chosen. Click "Next."
     
13. Add titles as desired and (under the "Legend" tab) uncheck the "Legend" box. (It’s useless and clutters the chart.) Click "Next." Place the chart as a new work sheet (titling is optional). Click "Next." Place the chart as a new sheet (titling is optional).

In the now created Scatter Plot chart, place your cursor on any point to see its interval value (e.g., kW) and the percent of intervals having values higher than that interval value. In most cases, the logical place to set a demand limiting goal is where the curve begins to level off (in our example, that's about 2261 kW). Any lower level involves many more hours of curtailment or generator operation, either of which may become problematic.

But When Did Those Peaks Occur?

To start a peak shaving program, it's helpful to understand what conditions typically cause high peaks. Doing so is simplified by knowing the dates and times when those peaks occurred, and the conditions that pertained at those times. The easiest way to determine these times and dates is to create a second chart of our interval data, this time also using the date and time information from the interval data table.

Using data in a calendar-style array (i.e., with the first column being the day/date, the top row being the interval's time, and the intersection points being the kW or kWh interval data), follow the procedure found at "Better 3-D Load Profiles," Tip of the Month for September 2003 (to be posted soon), but choose "3D column with a cylindrical shape" (instead of "Surface") as the chart type.

In the resulting 3-D chart, set the minimum value for the Z-axis at the level where demand control is desired (in our example, 2261 kW). Do so by right-clicking on the chart's Z-axis and choosing "Format Axis." Under the "Scale" tab that appears, input that value in the "Minimum" field. The 3-D chart will then show only those cylinders with interval data values (i.e., heights) above that minimum. Touch your cursor to those cylinders to see tags appear that contain the time, day of the week, date, and interval value (see Figure 2).

Note: If data is limited to less than an 8-month period, a "slice" chart will yield the same capability. The slice chart procedure can be found at "Better 3-D Load Profiles," Tip of the Month for September 2003 (to be posted soon).

With that information in hand, review and correlate outdoor temperature, building operating schedules, and other variables that could have influenced peak demand at those times. Doing so will provide an understanding of how to control loads next time those situations arise. Prepare a plan for adjusting loads (e.g., allowing some space temperatures to rise for short periods, reducing nonessential lighting, sequencing variable-speed fans, adding thermal storage) or supplying them without using utility power (e.g., running an on-site generator). Whenever possible, automate such systems or make EMS operators responsible for making appropriate adjustments when demand begins to approach a pre-established limit.

Energywiz, Inc.
Adding New Dimensions to Energy Services ^SM