Install the Four Pillars of a Quality Organization to Create Better Products

A company’s profitability is mostly affected by three factors: the ability to produce quality products consistently, the efficiency with which products are produced and the ability to sell the products. Total quality control has a vigorous impact on each of these factors, and, therefore, profitability.

Through careful analysis of customer wants and needs, a company can provide a product that attracts and motivates customers to make purchases, thus increasing salability. When the quality of the product design and production process is established with efficiency in mind, manufacturing costs can be substantially reduced. Taking these actions to improve production quality also helps to minimize the possibility of negative cost offsets such as recall action or liability suits.

Producing quality products requires a number of actions. Companies may find it helpful to organize these actions under the four pillars of quality: developmental quality, industrial quality, supplier quality and customer quality.

Product Creation Process

The product creation process for a great product or service involves five basic domains:

1. Functional – This domain consists of functional requirements, such as design goals or what the product or service must achieve to meet customer attributes from the viewpoint of the designer.

2. System – This domain consists of design parameters, including the physical characteristics or activities that are selected to meet functional goals.

3. Process – This domain consists of the process variables that make up the design parameters.

4. Subsystem – This domain consists of component-level requirements, such as the desired performance required from every critical component of the product.

5. Customer – This domain appears in the process twice (as shown in Figure 1). The first customer domain consists of customer attributes: a characterization of needs, wants or delights that define a successful product or service from a customer perspective. The last customer domain consists of customer complaints. A product characteristic that fails to meet customer expectation may trigger a customer complaint.

Figure 1: Product Creation Process

These domains that make up the product creation process fit into the four pillars of a quality organization (Figure 2). The four pillars and their roles are:

1. Development quality – Discover the customer needs to develop the product. This involves taking a proactive approach.

2. Industrial quality – Validate product design for manufacturability and develop the manufacturing process.

3. Supplier quality – Work closely with critical component suppliers to ensure desired quality level of these components.

4. Customer quality – Get early customer feedback and coordinate with different customer interfaces, such as call centers, service centers and online product support. This is a more reactive role.

Figure 2: Product Creation Process Within the Four Pillars of a Quality Organization

Certain activities need to be performed in each domain, and tracking these activities from one domain to another is critical. The better the communication between these domains, the better the design of the product.

Development Quality

Team members in the first pillar, development quality, must take the lead in driving the early project stages. The following activities are performed by the development quality team:

Collecting the voice of customer (VOC) – There are many approaches to capturing VOC, such as surveys, focus groups or interviews. It can be done by external agencies or by the internal project team. Once they have VOC, team members need to systematically convert the information to product performance characteristics, and then to product critical to quality (CTQ) parameters, with a defined target and specification.

For example, one call center found through collecting VOC that customers want a quick response to their queries. The team members converted this VOC to a measurable CTQ parameter (time) with a target (less than 5 minutes) and specification (15 minutes maximum).

Platform selection and architecture review – The product designer selects the correct platform. The development quality team takes the lead in reviewing it with the project team and validates it against the VOC.

Design failure mode and effects analysis (DFMEA) – Creating the DFMEA helps the team to identify the ways a product can fail to meet CTQs during design. Estimating the risks of specific causes with regard to failures and prioritizing actions for improvement is the main objective of this exercise.

Past learning and postmortem – It is always good to learn from past mistakes. Lessons learned from past projects and plans for mitigating any future mistakes are very useful for designing robust next-generation products.

Key component release – After releasing product design, the team should identify key components in the product design and release their detail specifications (functional as well as mechanical) to the purchase and supplier quality group to work with the supplier to attain them.

Customer and field trials – The team should ask a group of customers to test usability help validate whether the design addresses VOC, keeping in mind environment and operating conditions.

CTQ validation – They should also conduct further product testing against CTQs before releasing the design to confirm capability.

Product quality plan – This is a detailed product test plan for zero hour, primarily focused in the product performance domain. The team must clearly define what, how and when to test.

Industrial Quality

Organizational success lies in efficiently commercializing an innovation. The following activities, overseen by the industrial quality team, help meet that goal:

PFMEA and control plan – Once the product design is confirmed and released, the industrial quality team starts working on process design to ensure high productivity. They should create and regularly update a process failure mode and effects analysis (PFMEA) and control plan.

Work instruction and SOP management – The team must also detail work instructions and standard operating procedures (SOP) for the complete process.

CTQ deployment – This activity involves regularly measuring product CTQ characteristics during the production process.

Statistical process control – In cases where CTQs can’t be measured at the product level, they must be converted into critical-to-process parameters and should be monitored and controlled through statistical process control techniques.

8D management – This is a team-based problem solving methodology. Each discipline is involves a checklist of assessment questions, such as “What is wrong?” and “When, where, how much?” The 8D flowchart is displayed in Figure 3.

Figure 3: 8D Problem Solving Methodology

D1 – Select the team

D2 – Describe the problem

D3 – Implement and verify interim/containment actions

D4 – Define and verify root causes: Identify potential causes and select likely causes. If the potential cause is a root cause, identify possible corrective actions.

D5 – Verify corrective actions

D6 – Implement permanent corrective actions

D7 – Prevent recurrence

D8 – Recognize the team and/or individuals who worked on the problem

MSA – A gauge repeatability and reproducibility (GR&R) is the most popular technique for completing a measurement system analysis (MSA).

Product safety and traceability – The team should define a critical-to-safety checklist and follow it religiously. The team also should be able to trace critical components, process parameters and product test results. This data is useful in effectively managing defective product recalls in abnormal circumstances.

Life and reliability testing – The purpose of this testing is to re-confirm the probability of a given product to successfully perform a required function without failure, under specified environmental conditions, for a specified period of time. Product reliability is achieved by getting the product design right, and ensuring subsequent manufacturing conformity to that design.

Supplier Quality

Supplier quality is responsible for ensuring desired quality level and traceability of critical components identified by the development quality team during the product design. Their duties include:

Supplier selection – Team members should consider the following factors while selecting a supplier: product complexity and expertise in similar technology, geographic location, and long-term management commitment to quality and quantity.

Quality contracting – A quality contract with clear targets and compensation should be in place before starting business with a supplier.

Subsystem FMEA – A subsystem or component-level FMEA can help the team identify ways in which a module can fail to meet critical system-level customer requirements during the early stages of product design and in the manufacturing process. The team should estimate the risks related to these failures and prioritize actions for improvement.

Supplier competency development – The supplier quality role is entirely different from incoming quality control. The former’s role is more proactive and is mainly based on strategic and pattern-based thinking compared with incoming quality control, which is based on event-based thinking. The supplier quality team should work closely with the supplier to find out areas for improvement and develop the supplier through training and coaching.

Customer Quality

Although this pillar is more reactive than the other three, the customer quality team members play a key role in keeping customers happy after a product launches.

Define service scenario – During the design phase, the customer quality team must validate the product design for serviceability. This review helps organizations to maintain lower service costs during the product repairing process in the field. The product service scenario is defined during the design phase on the basis of product complexity and country cost structure, and whether the product should be replaced or repaired in the field.

Customer support and customer complaint addresser – This role consists of providing customer support, handling customer complaints and providing speedy solutions. Often, companies prefer online customer support, which is fast and cost effective. Some companies also offer online chatting to help solve customer complaints.

Customer satisfaction index management – Measuring the customer satisfaction index and putting an action plan in place is the most critical activity performed by the customer quality team. Corporate sectors are putting a lot of effort into improving their customer satisfaction index compared to their competitors because they are aware that even a single-digit increase in the customer satisfaction index can increase sales values significantly and lead to a healthy bottom line in the balance sheet.

Field quality improvement – This improvement is achieved through systematic analysis of returned products to find the root cause of the returns and put an action plan in place. Because the impact of these actions takes some time, the team must be very aggressive in execution.

Early customer feedback management – Due to rapid technology changes, sometimes a product’s lifecycle is not very long and by the time the customer quality team gets customer feedback, the product may not longer be in production. Early customer feedback is desirable in such cases.

Getting Quality Teams Involved

High-quality products and services help keep customers satisfied, and therefore are the key to survival for any enterprise. To have a better chance of beating their competition, organizations should involve quality teams in the entire product creation process, from conceiving the idea to delivering the product. The quality team should be able to help the organization:

1. Understand customer needs.

2. Provide high-quality and reliability consistent products and services.

3. Keep up with the pace of change (technological as well as political and social).

4. Be able topredict what the customer will want one year or 10 years from now.

Continue reading

Time-based Metrics: Accounting for the Differences

Green Belts and Black Belts produce time series plots of the primary Y metric during the measure phase of projects. The time dimension used to plot the data takes one of two forms, start-based or end-based, and the difference between the two can be substantial. Quality professionals risk incorrect conclusions by not understanding and accounting for this difference.

To understand the phenomenon, imagine transactions that take months to complete like selling homes. A sample data set might look like this:

Real Estate Data Set
Address	Appraisal Order Date	Sale Date	Appraised Value	Sale Price	Percent Difference
123 N. St.	1/1	6/1	100,000	93,000	-7%
234 S. St.	2/1	5/1	100,000	95,000	-5%
345 E. St.	3/1	3/1	100,000	97,000	-3%
456 W. St.	4/1	4/1	100,000	98,000	-2%

Note the difference in time series plots given the exact same data, but with different time dimensions. The project champion or process owner might come to a different conclusion depending on which graph he is shown.

Figure 1: Same Data but Different Time Dimensions

Start-based versus End-based Metrics

Typically, businesses prefer end-based metrics because performance is locked once the transactions are completed. Businesses prefer this because start-based metrics appear to change as older transactions complete their cycles. In the preceding example, the value for 123 North Street (-7 percent) would not be known until June and consequently appears to “change” the plotted value for January after the fact (i.e., going from a null value to -7 percent). Because financial performance is based on the business completed in a prior month or quarter, businesses almost always prefer end-based metrics.

The problem with end-based metrics is that if a project team implements an improvement on April 15th to reduce the percent difference metric it might appear that the project team made things worse.

Figure 2: Reducing the Percent Difference Metric

This is not the case since transactions subject to the improvement have not yet registered, but this might not be apparent to the uninformed.

Continue reading

The benefits of PRINCE2 Agile®

When one of the world’s most recognised project management certifications, PRINCE2®, announced its new agile sibling back in 2015 it caused quite a stir, and for good reason.

Bridging the gap between what is seen as a more ‘traditional’ project management methodology, and the techniques that make agile approaches so successful in their own right, PRINCE2 Agile® gives the best of both worlds. So what exactly are some of its benefits? Let’s break it down for you…

It’s adaptable

As it’s based on the PRINCE2 framework, PRINCE2 Agile can be applied to any project environment with ease, providing guidance that can be adapted and tailored depending on your industry sector, size of project, product type, the list goes on. It’s collaboratively built and can be used with any established agile approach, so your possibilities are well and truly endless.

It’s a good introduction to agile

If you’re already using PRINCE2 in your organisation, PRINCE2 Agile is a good way to start introducing agile techniques without, essentially, ripping everything out and starting over. Equally, there are no pre-requisites to sit the Foundation course, and if you already have PRINCE2 Practitioner then you can jump straight to the Practitioner level of PRINCE2 Agile – happy days!

It focuses on both the project and the outcome

While agile techniques are great for ensuring that you get the highest quality and best value out of your product at the delivery level, it can help to add in some structure to aid the management of the actual project itself. Ensuring that your projects adhere to the strategic objectives of your business is where PRINCE2’s framework comes in really handy.

It’s widely used

With PRINCE2 already being one of the most widely accepted project management methodologies in the world, getting a PRINCE2 Agile qualification will serve you in good stead. Whether your current organisation already uses PRINCE2, and they’re looking for a way to incorporate agile into their projects, or you’re looking for an agile certification that will help you in your future job prospects, you’re sure to reap the benefits of PRINCE2 Agile’s popularity.

You can start seeing value earlier

One of the main features that characterises agile projects is the collaborative, incremental style of delivery. Unlike some more traditional methods, where you may only be able to see the finished product or products at the end of the project, PRINCE2 Agile’s incremental delivery process allows you to start benefiting from the value of your product(s) in stages throughout the project.

It’s the perfect blend

If you wanted to, you could try to incorporate agile techniques into your projects without a qualification. However, you may find that the differing approaches often conflict with one another, having a detrimental effect on your projects. With PRINCE2 Agile, you’re taught how to get the best out of both approaches, with little to no conflict whatsoever – the dream!

Continue reading

Using Monte Carlo Simulation as Process Control Aid

Monte Carlo simulations are often used as a powerful tool in the Analyze or the Improve phase of a Six Sigma DMAIC (Define, Measure, Analyze, Improve, Control) project. For instance, Monte Carlo simulations can be used to improve the capability of processes. The Monte Carlo simulations can quantify the expected variation improvement. However, simulations are not only very powerful when utilized in the earlier phases of a DMAIC project, they also are a powerful tool in statistical process control.

Methodology

A project involving carbon nanotubes (CNTs) provides an example.

Carbon nanotubes have become the basis of intense research due to their intriguing mechanical and electrical properties. In Motorola’s Embedded Research Systems Laboratory CNT’s electrical conductivity and high aspect ratio (length in the m, diameters in the nm range) have been utilized in one particular application for field emitter displays. The adjacent image shows a 4.6-inch display where millions of CNTs have been selectively deposited on a glass cathode while their emission has been regulated for display function.

Such fully processed displays represent a considerable resource commitment due to their manufacturing complexity. Therefore, blanket, non-patterned, diode samples were initiated as test and optimization vehicles. CNT growth was achieved by a two-step deposition process – first a metallic catalyst is deposited in a separate tool, followed by the deposition of the CNTs in a chemical vapor deposition reactor. After deposition of the CNTs, the emission properties were investigated by extracting the emission current on an anode and exposing the CNTs to an electric field. The current density was an indicator of brightness, convoluted by the phosphorus efficacy of the anode.

The brightness of the display was maximized using a classical design of experiments (DOE) approach in what is called subsequently the “laboratory phase.” The most efficient method for optimization is a sequential approach that is partitioned into three separate components – screening DOE, path of steepest ascent (PoA) and response surface method (RSM), as illustrated in Figure1. Off-the-shelf software was used for the statistical analysis. The sequential experimental approach allowed for resource efficient optimization of the emission current density. Typically, this three-tiered approach is now used in Monte Carlo simulation to predict the process variability followed by laboratory verification runs. In the approach illustrated in Figure 1, it is demonstrated that the Monte Carlo simulation in what is termed the “simulation phase” also can be used as a process troubleshooting tool and may lead to discoveries of hidden influential factors. Simulation software was used to carry out the Monte Carlo simulation. The statistical model obtained from the laboratory phase was used as a transfer function to calculate the expected emission output current density. Since the control of the significant factors is limited, their variation led to a distribution of the resulting emission current density.

The DOE demonstrated that the output distribution is critically dependent on the input variation distributions. By systematically changing the significant factor distribution in a computer experiment, a good fit to the measured emission current density was obtained. Then the obtained response model can be analyzed for its sensitivity to the input distributions. The response model also can be used to detect significant deviations, for example, deviations caused by a process shift, from the expected output curve.

Figure 1: Flowchart of the Experimental Procedure

Laboratory Phase

A screening DOE (½ fraction of a 2-level, 4-factor resolution, IV, 24-1IV) was carried out in the initial phase to reveal what factors had significant influence on the emission current. The Pareto chart in Figure 2 identifies the two most significant factors, A and B, whereas D had only a minor influence. It also shows that some interaction between A and B was present and Factor C had no impact. The inset displays the residuals of the statistical model plotted against the most significant Factor A. The response surface was curved; therefore a RSM was required for true optimization.

Figure 2: Pareto Chart of the Standardized Effects

Figure 3: Contour Plot of Transformed Emission Current Density

Figure 4: Contour Plot of Emission Current Density (mA/cm2)

In the contour plot of Figure 3, an inverse square root transformation was used on the emission current density to normalize the data and plotted versus the two most significant factors – A and B. The analysis of variance (ANOVA) table is given in the inset. Results of Figure 3 indicated that in order to improve emission current density, Factor A should be increased and Factor B decreased. This course was pursued with the PoA. It rapidly became clear that while Factor A was easily controlled and increased, Factor B was not. Lowering Factor B below a certain threshold resulted in loss of controllability and a compromise had to be found. Therefore the most flexible and most important factor was Factor A. As Factor A was steadily increased a local maximum for emission current was found with the RSM centered on that new high point.

The results of the RSM are shown in Figure 4 with the ANOVA table as an inset. The contour map of the model obtained by the analysis indicates the region of maximized emission current. Figure 5 illustrates the effect of the optimization experiments. The boxplot summary of the emission current density is shown over time grouped by month. The current density is plotted on a logarithmic scale (each of the dashed lines representing one order of magnitude) and the median values for each month are indicated in bold numbers. It can easily be seen that the improvement equates to more than three orders of magnitude (from ca. -1.25 to ca. to +1.3, averaged over the last six months). The two images of the inset of Figure 5 illustrate the increase in brightness. The shutter speed of the camera had to be reduced by ca. 100x due to the much higher emission after the RSM. If the camera shutter speed had been left the same, the right image would be overexposed. Nonetheless, the improvement is not only noticeable in the brightness but also in the much higher density of emitters.

Figure 5: Box Plot of Emission Current Density Over Time

To this point the CNT growth project appeared to be a conventional Six Sigma DMAIC project with significant and quantifiable improvement. The Control phase started and for several months the emission stayed at a high level. However, in the fourth quarter of 2005 the process unexpectedly underwent a shift to lower emission. Since Factor A had been identified as most important and controllable factor, initial investigation focused on this factor, however, it was found that it was within the expected fluctuations. No clear explanation could be derived from the inherent variations of the other significant factors either.

After a more thorough investigation, the root cause was identified as an underlying factor relating to the catalyst thin film surface morphology that had not been measured directly. The characterization required to obtain the film surface morphology data on a regular basis had been excessively time- and labor-intensive. It took nearly two months of thorough experimentation and painstakingly characterization to bring the process back into control. The question became: Could this shift have been prevented? One obvious answer was that the correct parameter needed to be tracked, however, in this case the parameter was unknown. Could the shift of the process still have been prevented, and if so how?

Simulation Phase

In a typical DMAIC improvement project the process improvement is followed by some experimental verification runs followed by the capability analysis. The proposed method is to have the laboratory phase followed by a simulation phase to aid in the Control phase and to avoid costly process shifts even when hidden factors are important.

Optimized process conditions had been obtained and implemented, but how robust was the emission current density due to process input parameters, Factors A, B and D? If no variation to the optimized process parameters existed, the transfer function, obtained from the statistical model, could simply be used – taking into account the model error – to accurately calculate the expected emission current density output. As that is an idealized scenario even with the best controls in place, it becomes critically important to study the effect of process input fluctuations on the output.

No historical data for any of the factors existed. However, during the improvement project, data collection on the most significant factor, Factor A, was initiated (after some hardware modification on the tool). Since it took considerable time to collect enough data to be statistically viable, best guess estimation for all three significant factors, A, B and D had to be done. Normal distributions with standard deviations of 20 percent, 20 percent and 10 percent of the respective means were assigned to the three factors (the latter variable was more effectively controlled) and Monte Carlo simulation carried out. The resulting output emission current distribution was then compared to the measured data, as shown in the flowchart of Figure 1. The simulated and measured data were equally divided into equivalent bins and the difference for each bin calculated. The sum of squares of these bins was then used as the response in an optimization Monte Carlo RSM.

Figure 6: Overlaid Simulations with Varies Degrees of Variation of Factor A

Factor A was identified in the very first screening DOE as the most significant factor on the emission current density and Figure 6 shows several graphs overlaid with Factor A input variation from 5 percent to 50 percent. It can be seen that this variation has a substantial effect not only on the range but also on the shape of the current density distribution. The smallest variation in Factor A resulted in the sharpest peak with a very high kurtosis.

As the variation is increased it can be seen that the curve broadens considerably. However, the broadening disproportionately occurs at the upper end of the emission curve while no important effect is observed at the low end of the spectrum. Only the introduction of a much higher value than the experimental set point in combination with greater variation of Factor B, as seen in Figure 7 allowed for the extension of the emission current histogram towards zero emission and thus achieve a good fit to the measured data. In the figure, Factor B and its variation on emission current density curve came from 5000 simulation trials. Factor A variation was held constant at 50 percent.

Figure 7: Simulation Software Output Showing Influence of Factor B

Figure 8: Overlay Plot of Monte Carlo Simulation of Emission Current Density and Measured Data

The resulting final fit to the measured emission current density output distribution is shown in Figure 8. The fit was obtained with a 50 percent variation of the most significant Factor A, 78 percent variation of Factor B and Factor D stayed at 10 percent variation. Inset is a simulated run with the same amount of runs as the measured data.

The curve of measured data appear somewhat serrated compared to the fitted curve which may be explained by the limited number of 230 data points when compared to about 4 million simulated data points. The inset of Figure 8 shows the result of a simulation with only 230 runs (i.e., the equivalent number of measured data). The graph is very similar to the measured data, as it displays again a certain degree of unevenness and some bins that are not occupied. Therefore, the discretization of the data seems to be due to the low number of measured data relative to the simulation. As mentioned, during the improvement process the measuring of Factor A run-to-run variation was initiated. As illustrated in Figure 9, the measured Factor A variation has a standard deviation of 42 percent of the mean and agrees quite well with the 50 percent of simulated curve. Also shown are the Factor B and D variation.

Figure 9: Distributions of All Process Input Factors After Good Fit to Measured Data

The high Factor B variation was unexpected. The large fluctuation of Factor B was directly related to an underlying factor that dramatically decreased the emission current density. The morphology was crucially influenced by any fluctuations or drift in Factor B. The instrumentation that controlled Factor B was indicating no drift and the important influencing underlying factor was not known at the time, thus leading to the drift.

Continue reading

The benefits of PRINCE2

Whether the name ‘PRINCE2®’ means nothing to you – ‘Prince who?’, we hear you ask – or it’s mentioned so often that your interest is well and truly piqued, there’s a reason it’s one of the world’s most popular project management methodologies.

88% of project managers with a PRINCE2 qualification say that it has helped them in their career, and the project success rate for those holding a PRINCE2 certification is a massive 67%. While these numbers speak for themselves, in this blog we’ll be detailing some of the key ways in which PRINCE2 can benefit both projects and project managers alike.

It’s tailorable

PRINCE2’s framework can be tailored to any project environment. Whether it’s the size or type of project, the organisation’s individual needs or the industry sector, PRINCE2 is flexible enough to be adapted efficiently without losing any of its effectiveness, even in agile projects.

It’s widely recognised

There are over one million certified PRINCE2 professionals across the globe, and 59% of all project managers hold a PRINCE2 qualification2. With numbers that big staring you in the face, it’s hard to deny PRINCE2’s popularity. In fact, these days it’s often the case that employers will actively look for the certification when recruiting. With the methodology being so widespread, getting PRINCE2-qualified can not only increase your credibility as a project manager, but it can expand your network and opportunities worldwide.

It introduces a common vocabulary

By managing your projects under the PRINCE2 framework, you’ll be introducing an established vocabulary into your business, which will aid communication and help your projects run more efficiently. And another benefit of PRINCE2’s international renown is that this vocabulary is common amongst project managers and businesses worldwide, so understanding it can further boost your integrity and career opportunities.

It’s relevant at all stages of your career

Whether you’re just starting out in project management, or you’re a seasoned pro with years of experience, PRINCE2 is relevant to you. From helping you to structure and manage your first project with its emphasis on managing by stages, to improving the way your existing project team functions with its keen focus on roles and responsibilities, PRINCE2 is useful regardless of your experience level.

It’s compatible with other qualifications

The beauty of PRINCE2 is that you can combine it with any other project management qualification. By providing the technical framework from which you can structure your projects, it allows space for you to integrate softer skills and techniques from other certifications.

It’s product-focussed

This may seem an obvious one, but it’s important to highlight. PRINCE2’s focus on the product(s) of a project from the very beginning means that all your efforts are rooted in the end goal, rather than the process by which you get there. Ultimately, this will help your projects to run more smoothly, and it will help your project team to work more effectively.

Source: prince2.com

Continue reading

Perfect Your Predictions with Yield and Single-Use Reliability Modeling

Yield is a key parameter in most manufacturing processes – tied to financial results, delivery and quality as shown in Figure 1.

Figure 1: Yield Tied to Gross Margin and Delivery

Yield modeling allows the team to predict future yield and prioritize opportunities to improve yield. A yield model combines predicted yields for each step of the process into a predicted yield distribution for the entire manufacturing or assembly process.

Single-use reliability or mission reliability is analogous to yield modeling and refers to the overall probability of success, such as the success of a military mission, the launch of a satellite, or the probability of success for a medical procedure. The overall probability of success for the mission or procedure is a combination of the probabilities of success for each step in the mission or procedure, similar to steps in a manufacturing process.

By using either – or both – of these models, you can better prioritize your processes for improvement.

The probability of success for each step of a process, procedure or mission can range between 0 percent to 100 percent, so it can be modeled using a statistical distribution that ranges between 0 percent and 100 percent, such as a beta distribution. The overall probability for success of the process, mission or procedure also can range between 0 percent and 100 percent, and can likewise be represented by a beta distribution. Fortunately, combining the probabilities of success for each step by multiplying beta distributions for each step results in another beta distribution representing the probability of success for the mission, procedure or process.

Beta Distributions

Beta distributions, initially referred to as Pearson type 1 distributions, represent the family of distributions bounded by 0 percent and 100 percent. The beta distribution has two parameters, often referred to as alpha and beta. Insight into the meaning of alpha and beta parameters can be provided by analogy to the Bernoulli and the binomial discrete distributions.

On an individual basis, a part can pass or fail or a step in the process can experience success or failure; this can be represented by the Bernoulli distribution. If there is a set of five parts or five iterations of the process step, this can be represented by the counts of successes among five Bernoulli distributions; the distribution of the numbers of successes among the five Bernoulli distributions will give the same result as a binomial distribution for the counts of the number of successes among a set of five trials. Figure 2 shows the overlay of binomial distribution with probability of success per trial of 50 percent and five trials, compared to the sum of five Bernoulli distribution results each with a probability of success per trial of 50 percent.

Figure 2: Overlay of Binomial Distribution

If the number of trials is much larger than five, the information will move from discrete counts of successes divided by total attempts and approach a continuous parameter, the probability of success.

Binomial and Beta Probability Equations

By analogy to the related binomial distribution, two parameters of the beta distribution, alpha and beta, reflect the expected number of good parts and bad parts, respectively. The equations of binomial and beta probabilities are as follows.

Binomial probability:

Where s is the number of successes in n trials

Equation 1

Equation 2

So, if a step in the process has been tried n number of times, and has succeeded s times and failed f =  n – s times, the probability of success can be estimated by a beta distribution with an alpha parameter of (s + 1) and a beta parameter of (f + 1), described as beta(s + 1, f + 1). This approach provides a useful way to model the probability of success for an individual step in a manufacturing process or mission or procedure using beta distributions.

If the success of a step in the process is based on a continuous parameter rather than a discrete pass/fail parameter, the probability of success can also be converted to a beta distribution. A measure of goodness for the continuous, such as Cpk, z-score or yield can be used to estimate the probability of passing, p. However, estimating the two parameters for a beta distribution require two values, and the probability of passing, p, must be supplemented by a second value.

This second value can be the number of samples, n, or a value for n can be assumed to reflect the degree of uncertainty in the Cpk, z-score or yield from the parameter distribution. Using the previously referenced analogy between the beta and binomial distributions, the variance of yield or single use reliability is variance = p(1 – p)/n. Values for alpha and beta can be estimated from these equations based on the beta-binomial distribution analogy shown in Equations 3 and 4.

Alpha ≈ (p2-p3)/Variance – p

Equation 3

Beta ≈ alpha*(1 – p)/p

Equation 4

Values for the alpha and beta for the probability of success can be estimated for each step – whether based on the actual or predicted numbers of passes and fails for a discrete parameter or the Cpk or z-score or yield for a continuous parameter. An example of this summary is shown in the table below of yield modeling for five process steps, some continuous (Cpk) and some discrete (pass/fail).

Table 1: Yield Modeling

These values of alpha and beta for each step can be combined into overall probability of success for the overall process, corresponding to the overall yield for a manufacturing process or the probability of success for a procedure or mission, as shown in Figure 3. You can combine the values to use Monte Carlo simulation or another method based on the generation of system moments method. Both methods can provide sensitivity analysis that can help in prioritizing opportunities to improve yield or reliability, as shown in Table 2.

Figure 3: Combined Distribution for Yield or Single Use/Mission Reliability

Table 2: Average Yield and Variance Contribution from Each Step of the Process or Procedure

Since the combined distribution is represented by a beta distribution, a lower confidence limit for yield or single-use reliability can be established, as shown in Table 3.

Table 3: Yield or Single Use Reliability at 95% Confidence Using the Cumulative Beta Distribution Function
LSL confidence interval	95%
Yield at 0.95 confidence	82.9%

Continue reading

Leadership skills for the future

You may have felt tested in your role in the past few months, but rest assured, you are not alone! Collectively our management skills have been tried, pushed and developed in a crash course. Whilst there is no text book to answer “how to manage during a global pandemic”, you may have found yourself looking back to theory for how to best handle business disruption. You may have also looked to peers and even competitors for strategies and best practice approaches. However, now is the time to look within!

Take a moment to reflect and learn lessons from your own leadership skills during the coronavirus pandemic thus far. A pause of contemplation and self-reflection has the power to set you up to be a better leader going forward. Here are just some of the skills you may well have gained for the future:

Agility and adaptability

Whether you are well-versed in agile project management practices, or the concept is new to you, your agility will have undoubtedly been challenged by COVID-19. Extreme changes on a global level have affected business across the globe. As a manager this means your role will have been encompassed by change navigation. Adaptability has become vital, and there has been pressure for quick action like never before.

Going forward, so much will have been learned from this. It will surely impact the pace of your future decision-making, your ability to react and respond to bumps in the road on any project, and how to navigate through external changes.

Innovation

Innovation goes hand in hand with being agile and adaptable. As you decision-make, pivot and find solutions, you are innovating faster than ever before. Standing still has not been an option for companies during these unprecedented times, and so innovation has been crucial. Whether it has been repurposing existing products, relocating existing talents, finding new ways to deliver, or creating new products and services based on changing demands, or finding new markets.

Embrace this trend and become a team or company unafraid to try new things and continually reinvent, reimagine and create. Innovation will see your company thrive post lockdown. So, make a habit of being aware of shifts in the landscape and attentive to change. Become accustomed to seeing new opportunities and acting upon them.

The personal touch

A huge majority of us have gotten to know our neighbours more during lockdown, and there has been a real sense of community developing worldwide. People are looking out for one another, helping those who are struggling, and on the whole being more friendly and personable; we’re all in this together after all. This collaborative spirit is likely to have reached our work lives too.

As a manager in particular, you are likely to have checked in with your staff more than you would have done previously. In part, this is due to employees working remotely, but also because of the seismic changes which we have all had to deal with – it is always worth checking messages have been received and that individuals are on the right track. The takeaway from this being that you will likely know your team that little bit better on the other side of these testing times. Ensure that two-way communication and your personal touch continues, as it might just make you a better leader.

Positivity

It is key that we look for the positives to take from the COVID-19 situation. Being more personable, innovative and agile going forward is certain to benefit your work life. Further development of each of these traits and soft skills will be key to success. So, reflect on these takeaways, and others too - perhaps you have surprised yourself with how you’ve handled business communications throughout disruption, or ways in which you have lifted team morale. Beyond being prime examples for any future interviews, these success stories should give you the confidence to rise to new challenges. Establishing and advancing these skills could be fundamental in the future of your career as a leader.

Continue reading

Lean Fills Prescription for Change at Hospital Pharmacy

The process improvement journey at Jordan Hospital took a large step forward last year when the organization was introduced to Lean Six Sigma concepts as part of a departmental optimization effort for its internal pharmacy. Jordan Hospital is an acute care, 150-bed, not-for-profit community hospital, serving 12 towns in Plymouth and Barnstable counties in Massachusetts, USA.

Jordan Hospital’s pharmacy is currently undergoing several major changes to improve operations, reduce costs and better serve the needs of customers via the application of Lean principles. The pharmacy, led by John Leone, director of pharmacy, is following Lean’s primary tenets to reduce the time the supplier takes to deliver a customer’s order, and to eliminate wastes (and costs) throughout the system.

The objective for the hospital, according to Russ Averna, leader of the hospital’s Lean program and the vice president of human resources, is to apply Lean principles to improve operations, reduce costs, and better serve the needs of customers. Many other industries, including those that are service related, have adopted Lean approaches to increase their effectiveness and efficiency.

To optimize the transformation to using Lean, the hospital worked with consultants to introduce the change management tools to a broad audience of people from different areas of the hospital. The objective was to identify potential problems surrounding upcoming changes and create actionable plans to mitigate concerns and foster facility-wide acceptance.

Building Consensus for Change

A project team focused on increasing the clinical role of pharmacists to create a new vision of the “pharmacist of the future.” Another team worked on enhancing the communication of all changes within the pharmacy, and planning to optimize the future integration of the oncology pharmacy staff with the oncology department to further improve patient care and reduce drug delivery turnaround time.

In addition to these consensus-building events, the leadership also challenged the pharmacy team to improve the current performance inside its own walls where there were opportunities for improvement, specifically looking at the first-dose process and IV/chemo preparation.

The first step was to understand the current state of the process. This included the development of a value stream map (Figure 1) to identify where there was waste in the process. The value stream map highlighted non-value-added steps, including significant wait times during which medication was not being processed (waiting on quality checks or delivery runs) as well as extensive travel distances for the technicians who had to fetch some medications from outside the immediate work area.

In addition to opportunities to reduce the overall product lead-time by improving the cycle time, travel distance and flow patterns, there was an issue of physical space constraints. This was relevant because space in the pharmacy was extremely tight and congested, with people running into each other while filling patient orders.

Figure 1: Value Stream Map

Rapid Results with Kaizen

The next step was to prepare for a Kaizen improvement activity. Kaizen is a Japanese word that means change for the better, or more commonly, continuous improvement. A Kaizen is a hands-on burst of improvement activity in the actual workspace that occurs during a relatively short period of time – usually three or four days.

A team, which was comprised of those who perform the activities every day, brainstormed improvement approaches, and then tested them on the actual production or service line (referred to as “trystorming”). This method results in rapid recognition of which solutions will work and which will not. The Kaizen team also created plans to sustain the gains once improvements were made, and developed reports to leadership on its actions.

In preparation, management set clear goals for the Kaizen team. The goals focused on creating a standard work environment and implementing an improved replenishment system using Kanban, a Japanese word meaning “sign board” or “signal.” Pharmacy Director John Leone said, “Despite the fact that our environment will change a lot over the next 12 months, there are many areas where we need to improve the process to optimize future changes. We assembled a great team, and we had every expectation they could achieve the targets.”

The first activity was a half-day training session to familiarize the team and attending executives with the Lean process improvement method and the Lean tools that would be used during the week. After the training was completed, the Kaizen began. The initial activity centered on a 5S of the pharmacy (sort, set in order, shine, standardize and sustain). This included evaluating the purpose and value of all supplies and activities in the work areas. This step alone cleared a significant amount of “clutter” from the area, improving visual control and eliminating several safety hazards. The 5S activity freed up an entire rack where medication that was no longer used was being stored.

Figure 2: Pharmacy Area Before and After 5S Activity

Creating and Sustaining Change

The team then split up and focused on the “bug juice” preparation process and on the IV preparation process and replenishment of this area. As soon as the teams agreed on how to approach the individual problem, they went ahead and dove into the “trystorm” phase where they implemented their ideas. The bug juice team quickly identified and implemented a visual aid and a standard process that would increase communication and reduce the amount of “waste” produced due to communication and replenishment issues with the operating room team.

The IV prep team implemented a two-bin replenishment system close to the point where the medications were used. The two-bin system prevents overstocking of material, provides a mechanism to assure first-in-first-out inventory management and supports a much more just-in-time-driven replenishment.

In addition, the workspace for frozen IV preparation was moved from a congested area to a new location in the space that was recovered through the 5S activity. In the newly created space, workers could focus on the delivery quality of this task without being interrupted. The original IV area had no clear organization of inventory near by. The new IV area had point-of-use equipment and frequently used inventory visible.

Figure 3: The Original IV Area and the New IV Area

The benefits for this 3.5-day project included a reduction of approximately 51 miles per year in travel distance for the technicians, a cycle time reduction of about 102 hours per year, plus additional space created in the tight area.

Figure 4: Filling IV Label – Before and After

Also, the “bug juice” improvements will reduce the waste of medication, which, before, was calculated to about $50,000 per year. Additionally, several safety issues were resolved on the spot, making the environment a safer place to work in. The group then created a 30-day follow-up plan to fully implement those changes that could not be completed within the Kaizen event week to assure that full benefits could be calculated and sustained.

The success of the project was credited to the make-up and performance of team members, the support given to the team, the constructive feedback given to the rapid changes, and the methods used to ensure that changes are communicated effectively. “We really emphasized the people side during Kaizen and the other activities. Change is difficult, but making sure we gave every opportunity to participate was key to the positive results we achieved,” Vice President Averna noted

Additional factors included thorough preparation, clear objectives, management support, a can-do and creative mindset, continuous communication with all affected areas, standardization of the improvements, and routine monitoring and tracking of improvements following implementation.

Continue reading