Forecast your maintenance requirements

Written exclusively for PlantServices.com, this article by Howard Forman explains the goal of forecasting: To have the right material available at the right time, in the right place and quantity without tying up inventory dollars.

By Howard Forman

PlantServices.com

We've all heard the arguments about forecasting in a maintenance environment. The common question is "Who are you kidding? No one will be able to predict maintenance requirements. How will a forecast be useful when no one has any idea of tomorrow's workload?"

In reality, a forecast greatly simplifies deciding what, when and how much to order. Forecasting provides an opportunity to respond proactively and resolve material and scheduling problems. In fact, improved material planning, scheduling and supplier coordination raises department utilization, shrinks inventory levels and enhances customer service.

Forecasting defined
APICS, the Educational Society for Resource Management, defines forecasting as "the business function that attempts to predict the use of products so they can be purchased or manufactured in appropriate quantities in advance." A forecast is "an estimate of future demand. A forecast can be determined by mathematical means using historical data, it can be created subjectively by using estimates from informal sources, or it can represent a combination of both techniques."

Why forecast?
A company derives benefit from a maintenance forecast, especially companies operating in a run-to-failure maintenance mode, waiting for a breakdown before procuring material. The goal of maintenance forecasting is to have the right material available at the right time, in the right place, in the right quantity without tying up inventory dollars or tolerating material shortages. A maintenance department with a forecast better serves internal customers by anticipating maintenance requirements and having material available to meet these needs. The forecast helps to respond to emergencies and other scheduled maintenance activities quickly.

Maintenance forecasting is essential to maintaining a high level of operational readiness. Forecasting can help balance the cost of carrying high inventory levels on every item against numerous rush orders to support maintenance activities and its potential for causing downtime.

Forecast principles
Two basic forecast principles are there's no such thing as a reliable forecast and one should expect the forecast to be wrong. Tracking and using forecast bias (the difference between forecast and actual) to calculate a standard deviation for the error improves forecast effectiveness. Use the standard deviation to calculate safety stock levels that cushion demand variability. Forecast principles also state that current-period forecasts are more accurate than later periods. A general understanding of these basic principles should help users improve their ability to forecast.

The maintenance department forecast should help predict repair frequency on the basis of qualitative or quantitative measures. Qualitative forecasts reflect a person's intuition, make use of informed opinions and tend to be subjective. Qualitative forecasts are used in the absence of historical data. For example, a new machine won't have a repair history. The forecast will be based on the manufacturers' suggested maintenance cycle and parts list. The maintenance department also will forecast other material critical to the operation performance. Qualitative techniques include:

• Delphi method
• Market research
• Panel consensus
• Historical analogy

The quantitative technique uses historical demand data to calculate a future forecast. This technique assumes the demand pattern will be repeated in the future. For example, if the maintenance department used 1,000 gallons of oil each month, a quantitative forecast will project a demand similar to past demand. This logic remains valid as long as no structural changes appear. For example, replacing a machine with a new model alters the underlying structure of the forecast.

Quantitative techniques include:

• Historical time series
• Causal studies
• Simulation models

This quantitative technique will help to identify characteristics such as trend (demand change), seasonality (use varying regularly over time), randomness (one-time occurrences), or cyclicality (time interval).

Forecasting steps
The first step in producing a forecast is to capture monthly or weekly usage data going back one to three years. Forecast accuracy improves with the number of periods for which data is captured. Collect records such as maintenance work order number, material and quantity used; issue date, and reason code. In addition, demand activity may require user-specified adjustments or modifications to accommodate material substitutions, resource constraints, equipment replacement, cancelled work orders and the like. Failure to validate and correct demand data skews forecasted quantities.

The second step is to determine the number of maintenance items to include in the forecast. Large numbers are not practical. Use a stratification process to identify items to be forecasted based on cost, frequency of use or lead-time.

The third step is to produce a bill of material for repairs to identify the parent item, along with the components required to perform a preventive or scheduled repair. For example, the engine repair kit defines the material used historically to make the repair (see Table 1 below). Place these items on a repair bill-of-material, along with a defined replacement factor (percent of time the repair used the part). The engine repair kit then will be forecasted based on historical use.

Table 1.

Level	Item #	Description	Qty	UOM	Replacement    Factor
0	A12345	Engine Repair Kit
1	B23777	Gasket	1	Ea	1.00
1	B38739	Oil	0.5	LI	1.00
1	C48282	Core Assembly	1	Ea	0.25
1	B33388	Hardware Kit	1	Ea	0.40
1	D77654	Spray Paint Can	1	Ea	1.00