Reading...
Play button
Play button
Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
60 Cards in this Set
- Front
- Back
|
push system
|
every worker maximizes outputs
focuses on keeping work stations busy instead of using materials efficiently more defects TT increases as WIP increases causes line bottlenecks and unfinished goods hard to respond to special orders and needed due to TT |
|
|
pull system
|
controlled last operation, Kanban cards control WIP
controls max. WIP accumulating at bottlenecks keeps materials busy, not operators; operators do not work unless given signal to produce no slack in system if problem arises TT and WIP are less; faster reaction to defects and less oppurtunity to create them |
|
|
lean production
|
pull system throughout plant
attacks waste, exposes problems and bottlenecks, streamlines production requires employee participation, industrial engineering, continuing improvement, total quality control small lot sizes assumes a stable environment |
|
|
Ford
|
cut out useless parts
parts are designed so that they ca be easily made put into motion by Ohno |
|
|
toyota production system (TPS)
|
searches for and eliminates waste in value chain
views every activity as an operation applies waste reduction also, respect for peopl |
|
|
ways to eliminate waste
|
5s
group technology quality at the source JIT production kanban system minimized set up times uniform plant loading focused factory networks |
|
|
closed mitt
|
complexity
labor overproduction space energy defects materials inventory time transportation |
|
|
5 s
|
minimizing waste through neatness
sort simplify sweep standardize sustain critical for set-up reduction, pull systems, maitenence, inventory mgt. |
|
|
departmental specialization
|
plant layout can cause a lot of unneccessary movement
|
|
|
group technology cells
|
can reduce movement and improve product flow
|
|
|
just-in-time
|
only produce what is needed
opposite of just-in-case ideal lot size is one minimize transit time frequent small deliveries easier to spot quality issues requires discipline and good problem-solving suppliers or warehouses must be close requires high quality requires minimum kanbans |
|
|
inventory...
|
hides problems
|
|
|
kanban
|
signaling device to control the flow of material:
cards empty containers lights colored golf balls etc. |
|
|
problems with long set-up times
|
long production runs and lead time
large lots long lead times |
|
|
set-up reduction
|
focused efforts
problem solving flexible equiptment |
|
|
plant loading
|
uniform loading helps to reduce labor costs
|
|
|
focused factory networks
|
small specialized plants that limit the amt. of products produced
sometimes only one type of product per facility |
|
|
forecasting
|
tool used for predicting demand based on past info.
important because demand is uncertain can be used for strategic planning, finance and accounting, marketing, production and operations always wrong more accurate for groups or families of items more accurate for shorter time periods should include an error estimate no substitute for calculated demand only as good as the info. included in the forecast history is not a perfect predictor of the future |
|
|
what to consider when looking at demand data
|
trends
seasonality cyclical elements autocorrelation random variation |
|
|
qualitative
|
rely on subjective opinions from experts
|
|
|
quantative
|
rely on data and analytical techniques
|
|
|
grass roots
|
deriving future demand by asking person closest to the customer
|
|
|
market research
|
trying to identify customer habits and new products
|
|
|
panel consensus
|
deriving demand info. from synergy of panel of experts
|
|
|
historical analysis
|
identifying a similar mkt
|
|
|
delphi method
|
similar to panel consensus but with concealed identities
|
|
|
time series
|
predicts future demand based on past trends
|
|
|
causal relationship
|
uses statistical techniques to establish relationships b/t items and demand
|
|
|
simulation
|
uses randomness and nonlinear effects
|
|
|
how to choose a forecasting model
|
data availability
time horizon required accuracy required resources |
|
|
time series: moving average
|
uses last t periods to predict data for period t+1
can be simple or weighted assumption that most accurate prediction of future demand is a linear combination of past demand |
|
|
simple moving average
|
forecast for next period = actual sales from each period/number of periods used for data
5-month average: more smoothing 3-month average: more responsive to change |
|
|
weighted moving average
|
used when we want to give more importance to certain periods
all probability weights must add up to one ability to give more importance to recent data without losing the impact of past data we choose weights depending on the importance of data, and if there is seasonality |
|
|
time series: exponential smoothing
|
prediction of future depends mostly on recent observation and the error of most recent forecast
alpha: smoothing constant; denotes the importance of past error why we use it: - less space needed for data - accurate - easy to understand - not complex to calculate - simple accuracy tests if a is low: little reaction to observed differences if a is high: more reaction to observed differences |
|
|
time series: exponential smoothing with trend
|
regular exponential smoothing will always lag behind in trend
beta: trend-smoothing constant F=forecast without trend and T = trend component |
|
|
linear regression in forecasting
|
based on fitting data to a straight line and that change in one variable explains changes in others
dependent variable = a + b*independent variable we try to explore which independent variables affect the dependent variable error = observed - predicted goal: to minimize the sum of squared errors |
|
|
forecast error
|
errors can be:
biased random forecast error = difference between actual and forecasted value (aka residual) |
|
|
mean forecast error
|
based on bias
average error in observations more positive or negative answer means forecast is biased -- worse performance |
|
|
mean absolute deviation
|
average absolute error in observations
higher MAD implies worse performance if errors are normal dist., sigma = 1.25MAD |
|
|
low MFE and MAD
|
errors are small and unbaised
|
|
|
low MFE, high MAD
|
one average, arrows hit bullseye
|
|
|
high MFE and MAD
|
forecast is inaccurate and unbiased
|
|
|
how to measure forecasts for accuracy
|
by measuring mean absolute deviation or standard deviation of forecast error
|
|
|
tracking signal
|
helps measure accuracy of forecasts
measure of how often estimations are above or below actual values TS = RSFE/MAD positive tracking signal: actual values are usually above forecast values negative tracking signal: actual values are usually below forecasted values if TS>4 or less than <-4, investigate! |
|
|
inventory
|
raw materials, component parts, WIP, or finished goods held at a location in the supply chain
|
|
|
inventory at macro level
|
biggest corp. asset
investment is currently over $1.25 trillion accounts for almost 25% of GNP efficiency can be increased by controlling inventories |
|
|
inventory at firm level
|
sales growth = right inventory, at right place, at right time
cost reduction = less money tied up in inventory, inv. mgt., and obsalecence this can all lead to a higher profit |
|
|
what to consider with inventory
|
the cost of not having it
cost of getting more, cost of drawing money cost of holding and storing lost interest price discounts how much you consume safety against uncertainty |
|
|
costs of inventory
|
physical holding costs
oppurtunity costs operational costs |
|
|
benefits of inventory
|
hedge against uncertain supply and demand
economize ordering costs smoothing we should keep inventory to satisfy supply and demand in the most cost-effective way |
|
|
types of inventory models
|
multi-period
single-period |
|
|
multi-period model
|
how often review? when to place an order? how to order? how much to keep in stock?
retailer depends on supply and demand demand is determintistic = D/units per year known ordering cost (S) and immediate replenishment annual holding cost = H/unit purchasing cost = C/unit lets say we decide to order in batches of Q: # of periods = D/Q average inventory/period = Q/2 purchasing costs = D*C ordering cost = D/Q * S inventory cost = Q/2 * H economic order qty. = sqrt 2SD/H lead time = amt of time between when order is placed and when it is received we should order the EOQ and reorder at the ROP |
|
|
safety stock
|
there in case of a stockout
= z * std. deviation in LT demand |
|
|
normal curve w/ service level
|
service level = mean
|
|
|
caution: std. deviation in LT demand
|
std. deviation for multiple periods is the sqrt. of multiple variances, not the sum of multiple std. deviations
|
|
|
average inventory
|
= order qty./2 + safety stock
|
|
|
newsvendor framework
|
one chance to decide stocking qty. for product
demand is uncertain known marginal profit for each unit sold and known marginal loss for each unit not sold goal: to maximize expected profit examples: perishable goods, short selling season buy another unit only if the expected new profit of doing so is positive |
|
|
variables in newsvendor
|
c: cost
r: selling price s: salvage value MP: marg. profit in selling a stocked unit (r-c) ML: marg. loss from NOT selling a stocked unit (c-s) x: number of newspapers you buy P(X): prob. that the xth newspaper is sold |
|
|
newsvendor with normal distribution
|
Pc = ML/MP+ML
use table to find z value for 1-Pc x* = mean + z*std. deviation |
|
|
what we learn from the newsvendor model
|
forecasts are always wrong -- demand estimate that just gives you the mean is too simple; you also need the std.deviation
optimal order qty. depends on the cost of stocking too much or stocking too little the smaller the SD, the closer the order will be to the mean |
