| |
Technology Development
Over the last 20 years, PredictionProbe' top engineers have developed
leading-edge predictive technologies in a variety of areas. These technologies
are rapidly changing the way companies do business. Some of these technologies
are:
- Probabilistic Design / Analysis
- Sensitivity Analysis
- Mathematical Modeling of Complex Processes
- Distributional / Uncertainty Analysis
- Error Bounding
- Competitive Analysis
- Process Optimization
Examples of Successes Using Probabilistic Methods
As companies in many fields are discovering that it is better to understand
how Variation impacts their products and to exploit this variation for
competitive purposes, Probabilistic Methods have exploded. Probabilistic
Methods may be the next wave (following the implementation of downsizing,
outsourcing, offshore manufacturing, and offshore professional sourcing) to
achieve that edge over the competition.
Examples with Engineering have included:
- Weight Savings
- Reliability Improvement
- Cost Reduction
- Life Extension
- Root-Cause Identification
- Uncertainty / Risk / Safety Quantification (Reliability Prediction)
Specific Applications of Probabilistic Methods
- Designers at an aerospace company wanted to try to reduce the weight of
a Turbine Disk design since it was felt that plenty of margin existed. The
Low Cycle Fatigue (LCF) lives at various disk locations were calculated as
a function of variations in temperature, rotor speed, pressure, material
properties, blade loads, snap fits, and dimensions.
The Probabilistic Analysis showed excessive conservatism over the
deterministic (standard engineering design) approach. Weight was
systematically removed until most of the excess conservatism was removed.
In the 100 lb disk, approximately 8 lbs. of weight was removed. The cost
savings in raw material and performance over the life cycle of a fleet of
engines was $90M.
- For a Fan Disk Application at a large manufacturing company, based on
the excess conservatism that existed in a design as shown by the
probabilistic analysis, the company was able to extend the life to twice
its deterministically predicted life. This allowed the company (which
warranties the disk for a specified lifetime and replaces it with a new
disk at that time) to make an additional $58M on the program.
- For a new power generator program, one company determined the offering
price which maximized the expected profit that it would receive by
conducting a Probabilistic Competitive Analysis. Maximizing expected
profit is the same as simultaneously maximizing both profit and win
probability. The Competitive Analysis took into account uncertainties in
the competitor's bid price as well as uncertainties in win probability vs.
how much the offered price was above / below the competitor's bid. Based
on this analysis, the company won a $200M contract.
- A government agency was able to predict the distribution of reliability
for an entire large vehicle system by using Probabilistic Methods. This
was accomplished by developing probabilistic models for each component on
the large vehicle system and aggregating them to an overall model. Now
that the model has been developed, the government agency can quickly
determine what the effect of a change in any one component will have on
the overall large vehicle system reliability. Trade-offs between Cost and
Reliability can now be made.
- A casting manufacturing company was able to use probabilistic methods to
reduce the scrap rate due to surface blemishes of its product by reducing
the rate from 42% to less than 1%. Based on the production schedules, this
resulted in savings of $15M per year.
- An aerospace company was having a vibration problem in one of its
aircraft. The solution to the problem was to replace the entire engine
with a new one, and the problem went away. However, they did not
understand the root cause, and the fix was extremely expensive. Instead, a
probabilistic analysis was conducted to investigate the possible causes of
variation. After systematically sampling about a dozen engines and running
a probabilistic analysis, the component where the problem originated was
immediately isolated. Once this root cause was identified, it was only
necessary to replace the small component - not the entire engine. The
resulting cost of the fix was only $1000 as opposed to the original cost
of the engine replacement of $2.5M.
|
