DMADV - Analyze

PD team members use the Analyze Phase of the DMADV mode to generate and investigate alternative high-level design concepts for the critical parameters (CTQsa dn/or CTPs), with nominal values and tolerances, developed in the Measure Phase. Further, the set of alternative designs are compared and studied to identify the best design to move forward into the Design Phase of the DMADV model.

Steps of the analyze phase

1 Generate alternative high-level design concepts for each CTQ and high-level CTP. a. Determine the level of difficulty of generating the design concept. * Why??? a. Use creative methods to generate alternative design concepts, if appropriate. a. Employ the theory of inventive problem solving (called TRIZ), which utilizes physics, chemistry, and mathematics to generate alternative design concepts, if appropriate. a. Use any other methods to generate alternative design concepts, if appropriate. 1 Reduce the number of alternative high-level design concepts for each CTQ and high-level CTP. a.The team members select viable design concepts for each critical parameter. a. The team members develop high-level designs for two or three of the viable design concepts. a. The team members assess the risks relating to the "best high-level" design concepts. a. The team members conduct a risk/hazard analysis (hazards and harms) for the "best high-level" design concepts using one or more of the following techniques: * FMEA * Hazard Analysis * Fault Tree Analysis (FTA) * Basic Reliability Concepts to quantify the risks on a FTA a. The team members optimize the Total Life Cycle Costs for the "best high-level" design concepts. a. The team members develop a process model (e.g., flowchart, blueprint, etc.) of the "best high-level" design concepts. a. The team members perform a discrete event simulation of the "best high-level" design to study its performance, if appropriate. 1 Combine the best parts of the remaining design concepts to create two or three composite best design concepts for each CTQ and high-level CTP. 1 Study the resulting composite design concepts. 1 Select the best design concept for each CTQ and high-level CTP.

Inputs of the analyze phase

• Prioritized list of CTQs, part features, service steps (high-level CTPs), etc.
• Nominal values and specification limits for CTQs, part features, service steps (highlevel CTPs), etc.
• Design scorecard for CTQs, part features, service steps (high-level CTPs), etc.
• Report on the status of existing patents and trademarks.

Generate high-level design concepts for critical parameters

effort level needed to create design concepts

$ Simple: Simple fairly obvious design concepts that require absolutely no invention. $ Small: System modifications or small improvements, with no invention(Git06, ch6.3). * An off-the-shelf design concept does not exist; the original design concept is being improved and slightly changed. These design concepts have contradictions that require tradeoffs and compromises between the components of the design concept. $ Inventive: An inventive design concept or a major change to an existing design concept (invention within the current paradigm) * These designs require a major change to an existing design concept because a well-known available design concept does not exist. Design concept contradictions do exist and require tradeoffs and compromises between the components of the design concept. Designers must study hundreds of design concept variants inside and outside of their discipline to generate a design concept for the critical parameter $ Radical: A significant inventive design concept or a radical change to an existing design concept (invention outside the current paradigm) * These designs require a new paradigm to create the critical design parameters. Design concept contradictions do exist and require tradeoffs/compromises between the components of the design concept. The original design concept is entirely changed, but the critical design parameters will perform the same function as the original critical parameter. * PD team members must study thousands of design concept variants outside their discipline to generate a design concept for the critical design parameters. $ Discovery: A new phenomenon or discovery (an entirely new invention) * These designs require a new discovery (beyond current science) used to create the critical parameter; this discovery can create an entirely new industry. PD team members must study tens or hundreds of thousands of design concept variants outside current knowledge to generate a design concept for the critical design parameters.

Breaking Down CTQs into CTPs

use a systematic (tree) diagram to decompose each CTQ into its component parts or functions, called critical-to-product characteristics or critical-to-process characteristics (high-level CTPs).

How do we actually get there? When do you start to look at the process? is that in the measurement phase? The Kano thing?

Using Thinking Habits and Tools for Generating Design Concepts

Lateral thinking is used to move from established ideas and perceptions to new ideas and perceptions. It is required for creating the ideas and perceptions necessary to generate design concepts.

Thinking habits

de Bono has developed the thinking habits required for creative thinking: focus and purpose, forward and parallel, perception and logic, values, and outcome and conclusion.

$ focus and purpose: keeps a thinker aimed in the right direction. It stops drift, confusion, and inefficiency in thinking about a topic. * What am I looking at (thinking about) right now? * What am I trying to do? $ forward and parallel: helps a thinker identify the next step in his or her thinking process. Is the next step forward or sideways? * Forward—So what follows? * Sideways—What else might there be? $ perception and logic: helps the thinker see his or her world (perception) and how to utilize his or her perceptions (logic) about the world. * Perception (Breadth)—How broad of a view am I taking? * Perception (Change)—In what other ways is it possible to look at things? * Logic—What follows from this? $ values: determines the value of the thinking to real life. * What are the values involved? * Who are affected by these values? $ outcome and conclusion: assists the thinker in harvesting the fruit of the thinking effort and feeling achievement in the outcome. It comes at the end of a thinking effort. *If you have not succeeded in reaching a conclusion: * What have I found out? * What is the sticking point? * If the thinker has succeeded in reaching a conclusion: * What is my answer? * Why do I think my answer will work?

Thinking tools

de Bono has developed several tools to assist individuals or teams to think creatively about a problem. e.g. ConsiderAllFactors.

Investigate alternative design concepts for each critical parameter

Reducing the set of potential design concepts

reduce the list of potential concepts to no more than six concepts.(Git06, ch6.4)

1 eliminate all "show-stopper" concepts. * These are concepts that have some extremely negative aspect associated with them. * likely to increase accidents, * create political problems * very expensive * break laws * are particularly unattractive. 1 team members can attempt to combine the concepts from different designs to create fewer designs with better features. * For example, team members may combine; simulation software, statistical software, and spreadsheet software to create an improved Six Sigma software package. 1 eliminate concepts that don’t fit with the organization’s * mission * values * beliefs * strategic objectives * or create stakeholder dissatisfaction.

identify the criteria upon which they will base their decisions concerning which design concepts to drop and which to carry forward into the next phase of analysis.

• Benefit/cost of concept
• Time required to realize the concept
• Organization’s ability to realize concept
• Effect of concept on organization’s strategy
• Legal/regulatory impact of concept
• Safety impact of concept on stakeholders
• Political ramifications of concept

establish importance weights for the design criteria using the following weighting scale:

• 1 = Very unimportant
• 2 = Unimportant
• 3 = Neutral
• 4 = Important
• 5 = Very important

select the "best" design concept using a PughMatrix

Steps for Developing a Process Model of a Design

There are nine steps for developing a process model of a design concept:

1 Identify the idea, object, or process for the design. 1 Identify the start and stop points, and any other boundaries. 1 Select the appropriate type of model (e.g., static or dynamic, etc.). 1 Construct initial high-level model. 1 Construct detailed-level model. 1 Validate the detailed-level model. 1 Optimize the model. 1 Perform a pilot test of the model. 1 Transfer the model to its proper user.

Using the symbols in Figure 6.2 standardizes process definition and documentation. Example of a Flowchart. Figure 6.3 shows a flowchart of a service design process. The process begins with team members developing the design of a trial service prototype, including specifications for the CTQs and high-level CTPs. Next, team members evaluate the prototype. A "bad" evaluation results in looping back to the design of the trial prototype specifications step for redesign of the service. A "good" evaluation results in a trial of the service. Next, team members evaluate the success of the trial service. A "bad" evaluation results in looping back to the design of the trail prototype specifications step. A "good" evaluation results in acceptance of the service design and the end of the service design process for the service under study.

Integrated Flowcharts (also called Deployment Flowcharts) An integrated flowchart is a special type of flowchart that is used to highlight responsibility for the various steps and decisions in a process. A small circle on an integrated flowchart indicates a stakeholder of the process that must be kept informed about a certain step or decision in the process. (Git06, p190)

Value-Added/Non-Value-Added Flowcharts (VA/NVA)

A VA/NVA flowchart, also called an opportunity flowchart, is a special type of integrated flowchart in which there are only two columns: a value-added column and a non-value-added column. The value-added column contains the steps and decisions of a product, service, or process that customers are willing to pay for because they positively change the product, service, or process in the view of the customer. The non-value-added column contains the steps and decisions of a product, service, or process that: 1 Customers are not willing to pay for. 1 Do not change the product or service. 1 Contain errors, defects, or omissions. 1 Require preparation or setup. 1 Involve control or inspection. 1 Involve over-production, special processing, and inventory. 1 Involve waiting and delays. p191

Value Analysis Matrix

A value analysis matrix is a tool that assists in identifying the steps and decisions of a product, service, or process that are non-value added, and the nature of the non-value-added steps or decisions. p192

Layout Flowchart

A layout flowchart depicts the floor plan of an area, usually including the flow of paperwork or goods and the location of equipment, file cabinets, storage areas, and so on. These flowcharts are especially helpful in visualizing and improving the layout to more efficiently utilize a space.

Dynamic Process Models of a Design

Definition of Simulation Simulation is a method for developing a model to numerically study and describe the alternative characteristics of an idea, product, service, or process over a pre-established time frame to select the best (optimal) alternative from the set of available alternatives.

!!! Simulation is discussed in detail in Chapter 13, "Discrete Event Simulation Models."

A quantitative statement of a simulation optimization model is: Simultaneously optimize {CTQa,CTQb,CTQc,CTQd,CTQe}, where

• CTQa[center, spread, shape] = f (X1a[center, spread, shape], …,Xna[center, spread, shape])
• CTQb[center, spread, shape] = f (X1b[center, spread, shape], ,…Xnb[center, spread, shape])
• CTQc[center, spread, shape] = f (X1c[center, spread, shape], …,Xnc[center, spread, shape])
• CTQd[center, spread, shape] = f (X1d[center, spread, shape], …Xnd[center, spread, shape])
• CTQe[center, spread, shape] = f (X1e[center, spread, shape], …,Xne[center, spread, shape])

Advantages and Disadvantages of Simulation The four main advantages of a simulation model are:

• Provides a method to study an extremely complex system.

• Allows experimentation on a system without disrupting the system.

• Promotes "what-if" analysis of a system with instant feedback.

• Conserves raw materials and resources. The four major disadvantages of a simulation model are:

• Can consume much time and resources.

• Is only as reliable as the assumptions and data utilized in building the model.

• Provides only estimates of true system parameters.

• Provides no guarantee of the optimality of the result.

• The first taxonomy is a static model versus a dynamic model. $ A static model: is one in which events are studied for only one point in time and, if repeated, are independent of time frames. $ A dynamic model: is one in which events are run over time.

• The second taxonomy is a deterministic model versus a probabilistic model. $ A deterministic model: assumes constant values for each independent parameter of a set of one or more independent variables; * for example, cycle time for steps in the process, which will yield a fixed and "determined" value of an outcome each and every time the independent variables are set at the same constant values. $ A probabilistic: model exhibits a distribution for each parameter; * for example, a stable and normally distributed cycle time with a mean of 30 minutes and a standard deviation of 5 minutes.

• The third taxonomy is a continuous model versus a discrete model. $ A continuous model: continuously runs a clock representing time in which there is randomness of occurrence over all events. $ A discrete model: advances a clock discretely according to the occurrence of the next event.

• the fourth taxonomy is a Monte Carlo simulation model versus a discrete event simulation model. $ Monte Carlo simulation: advances time by a clock. $ A discrete event: simulation advances time by events.

Performing a discrete event simulation

1 step 1 a. specify the idea, product, service, or process to be simulated, as well as the purpose of the simulation. a. collect input from relevant stakeholders about the simulation concerning their perspective on the parameters needed for the simulation. a. assign roles and responsibilities for building and experimenting with the discrete event simulation model. * This step requires that team members prepare a Gantt chart for the simulation study 1 Step 2 a. develop a conceptual model of the idea, product, service, or process. * Specify the CTQs. The CTQ was established in the Measure Phase of a Six Sigma project a. study the real-world idea, product, service, or process. * e.g. mapping out the existing shuttle bus routes and stops a. identify the CTPs. The CTPs include the following: $ Item (entity or transaction) being processed: are introduced into a model using a generator. A generator issues items with attributes (individualized characteristics) at a specified rate and quantity. * E.g.:In the shuttle bus example, there are two types of items: shuttle buses and passengers. Shuttle buses are generated (enter the system) according to an arrival distribution for each shuttle bus stop, a service (time to load up or wait) distribution for each shuttle bus stop, and a travel time distribution between each connected pair of shuttle bus stops. Passengers are generated (enter the system) according to an arrival distribution for each shuttle bus stop. $ Activity being performed on the item: in the system. * E.g.: In the shuttle bus example, the activities are advancing each shuttle bus from stop to stop through its route. $ Resources needed to process the item: as it passes through a system. * Number of shuttle buses. * Capacity of each shuttle bus. * Hours of operation for each shuttle bus (bus, driver, and dispatcher). $ Connections between the activities being performed on the items: links between the activities in a system. An entity enters and exits an activity through a connection line. Entities can split and reform while moving through connection lines (e.g., a three-part form is separated at one point in a process and reassembled at the end of the process). $ Properties (numerical characteristics) of the activities, entities, or resources: * In the shuttle bus example, one property of a passenger (item) is if the passenger is handicapped and requires wheelchair access to the shuttle bus (Yes = 1 or No = 0). * One property of a shuttle bus (item) is if the shuttle bus is equipped with wheelchair access (Yes = 1 or No = 0) * the number of seats on the shuttle bus (n). $ Control logic (rules) by which items are processed through activities using resources: Control logic defines the rules by which items are processed through the activities in a system using the resources that help define the system. * one instance of control logic is that a shuttle bus cannot move from a stop until a dispatcher knows the previous shuttle bus has left its stop. This prevents multiple shuttle buses at a given stop. a. develop a realistic model of the system under study that relates the CTPs to the CTQs; that is, to make a flowchart of the connections between activities of the system under study. * This requires that team members create a description or graphic (e.g., flowchart) of the system under study that indicates how the items flow through the system. In the shuttle bus example, this involves developing a discrete event simulation model of the shuttle bus system using a simulation software package. a. compare the flows of the items from the model with flows from the real-world system to determine if the simulated system reflects reality. * e.g: comparing the waiting time distributions for the passengers at each shuttle bus stop for each simulated and actual shuttle bus stop. a. revise the simulation model, if necessary, until its functioning adequately reflects the functioning of the actual system. * e.g: comparing the waiting time distributions for the passengers at each shuttle bus stop for each simulated and actual shuttle bus stop. a. revise the simulation model, if necessary, until its functioning adequately reflects the functioning of the actual system * e.g: revised until the waiting time distributions for passengers at each shuttle bus stop were not statistically different between the simulated system and the actual system. 1 collect the data needed to define the parameters of the simulation model; for example, arrival and service time distributions, and attributes of the items. a. collect data on the arrival patterns for the items (entities) in each activity * * How many items enter a given activity per unit of time? How many passengers arrive at each shuttle bus stop every 5 minutes? * Is the rate constant or variable over time? The rate is variable with a mean of 9 passengers with a standard deviation of 3 passengers. We can identify this as one assumption that is not likely true. Why? Academic class schedules vary throughout the day, and consequently, so do the number of students and faculty. Also, separately, the time to arrival between bus stops is also likely to vary with the time of day due to traffic conditions. * If the rate is variable, is it stable? The rate is a stable Poisson distribution [see Reference 6]. * Are the units batched (collect all ballots until they are processed)? Passengers are collected until the shuttle bus comes to the stop. a. estimate the processing times and resources (e.g., costs) for the items in each activity. * How long does it take to process an item in a given activity? This is the mean, standard deviation, and shape of the service time distribution, for a shuttle bus to load passengers at a stop. * Are processing times stable distributions? The service time distributions for all shuttle stops are stable distributions. * How much does it cost to process an entity in a given activity? It costs an average of $1.00 to process one passenger1 for the shuttle bus system. * What resources are used to process an entity by a given activity? In this example, these include shuttle buses, drivers, and dispatchers, to name a few resources. a. estimate storage and wait time statistics for items as they pass through the system * How long is an entity stored in a given activity? * Are the storage times constant or variable? * Are the entities batched? a. determine the resources required by each activity. * What resources are required to process entities by each activity? Shuttle buses, drivers, and dispatchers are required to move passengers around the campus. * What is the availability schedule for each resource for each activity? a. determine the control logic of the system. * What path does a particular entity take through the model? Passengers can select either the green, orange, or express lines. * What percentage of the entities takes path A1 versus path A2 through the model? Each route has an identified percentage of the passengers selecting it. * To summarize, data for all the CTPs specified above were collected for the simulation model of the shuttle bus system. 1 team members develop a computer model for the system under study using a discrete event simulation software package * the computer model includes: entities, activities, resources, control logic, and properties of items. 1 team members run and validate the simulation model against the real system to determine its validity as a surrogate for the actual system * In the shuttle bus system, the CTQs and CTPs from the simulation model compared favorably with the statistics of the real system. 1 team members plan one or more experiments to identify the optimal configuration of the system (best levels for the CTPs to obtain the best statistics for the CTQs). 1 run, record, and analyze the experimental alternatives using the computer simulation model. 1 study the results and select the best highlevel route design. 1 make recommendations concerning the best highlevel design concept by: * Documenting assumptions concerning the best high-level design concept. * Preparing report and presentation. * Avoiding the use of technical jargon.

Transfer High level design to process owner with design scorecards

Design scorecards help (Git06, 203):

• Establish nominal values and specification limits for each CTQ and CTP, as well as the target for the process sigma (Part A of the scorecard).
• Predict output of the Voice of the Process in respect to stability, shape, mean, and standard deviation, as well as DPU and predicted process sigma (Part B of the scorecard).
• Highlight problems and risks of CTQs and CTPs with respect to failure to meet desired process sigma levels.
• Track CTQs and CTPs throughout the entire life of the product, service, or process. | Scorecard -- Part A ||||| Scorecard -- Part B |||||| | (Voice of the Customer) ||||| (Predicted Voice of the Process |||||| | CTQ/CTP | Target | LSL | USL | Sigma trgt | Stable Y/N | Shape | Mean | std dev | DPU | PPS | | | | | | | | | | | | |

$ Target: Nominal Target. $ PPS: Predicted Process Sigma.